[{"data":1,"prerenderedAt":1769},["ShallowReactive",2],{"active-banner":3,"navbar-featured-partner-blog":24,"blog-\u002Fblog\u002Fa-comparison-of-transaction-buffer-snapshot-strategies-in-apache-pulsar":306,"navbar-pricing-featured":943,"blog-authors-\u002Fblog\u002Fa-comparison-of-transaction-buffer-snapshot-strategies-in-apache-pulsar":1713,"related-\u002Fblog\u002Fa-comparison-of-transaction-buffer-snapshot-strategies-in-apache-pulsar":1749},{"id":4,"title":5,"date":6,"dismissible":7,"extension":8,"link":9,"link2":10,"linkText":11,"linkText2":12,"meta":13,"stem":21,"variant":22,"__hash__":23},"banners\u002Fbanners\u002Flakestream-ufk-launch.md","StreamNative Introduces Lakestream Architecture and Launches Native Kafka Service","2026-04-07",true,"md","\u002Fblog\u002Ffrom-streams-to-lakestreams","https:\u002F\u002Fconsole.streamnative.cloud\u002Fsignup?from=banner_lakestream-launch","Read Announcement","Sign Up Now",{"body":14},{"type":15,"value":16,"toc":17},"minimark",[],{"title":18,"searchDepth":19,"depth":19,"links":20},"",2,[],"banners\u002Flakestream-ufk-launch","default","zRueBGutATZB0ZnFFHwaEV7F0Di4tnZUHhgOiI4cu6k",{"id":25,"title":26,"authors":27,"body":29,"category":289,"createdAt":290,"date":291,"description":292,"extension":8,"featured":7,"image":293,"isDraft":294,"link":290,"meta":295,"navigation":7,"order":296,"path":297,"readingTime":298,"relatedResources":290,"seo":299,"stem":300,"tags":301,"__hash__":305},"blogs\u002Fblog\u002Fstreamnative-recognized-in-the-forrester-wave-streaming-data-platforms-2025.md","StreamNative Recognized as a Contender in The Forrester Wave™: Streaming Data Platforms, Q4 2025",[28],"David Kjerrumgaard",{"type":15,"value":30,"toc":276},[31,39,47,51,67,73,78,81,87,102,109,115,118,124,127,134,140,143,146,157,163,169,172,175,178,184,191,194,197,204,207,210,224,229,233,237,241,245,249,251,268,270],[32,33,35],"h3",{"id":34},"receives-highest-possible-scores-in-both-the-messaging-and-resource-optimization-criteria",[36,37,38],"em",{},"Receives Highest Possible Scores in BOTH the Messaging and Resource Optimization Criteria",[40,41,43],"h2",{"id":42},"introduction",[44,45,46],"strong",{},"Introduction",[48,49,50],"p",{},"Real-time data has become the backbone of modern innovation. As artificial intelligence (AI) and digital services demand instantaneous insights, organizations are realizing that streaming data is no longer optional – it's essential for delivering timely, context-rich experiences. StreamNative's data streaming platform is built precisely for this reality, ensuring data is immediate, reliable, and ready to power critical applications.",[48,52,53,54,63,64],{},"Today, we're excited to announce that Forrester Research has named StreamNative as a Contender in its evaluation, ",[55,56,58],"a",{"href":57},"\u002Freports\u002Frecognized-in-the-forrester-wave-tm-streaming-data-platforms-q4-2025",[36,59,60],{},[44,61,62],{},"The Forrester Wave™: Streaming Data Platforms, Q4 2025",". This report evaluated 15 top streaming data platform providers, and we're proud to share that ",[44,65,66],{},"StreamNative received the highest scores possible—5 out of 5—in both the Messaging and Resource Optimization criteria.",[48,68,69,70],{},"***Forrester's Take: ***",[36,71,72],{},"\"StreamNative is a good fit for enterprises that want an Apache Pulsar implementation that is also compatible with Kafka APIs.\"",[48,74,75],{},[36,76,77],{},"— The Forrester Wave™: Streaming Data Platforms, Q4 2025",[48,79,80],{},"Being recognized in the Forrester Wave is a proud milestone, and for us, it highlights how far StreamNative has come in enabling enterprises to unlock the power of real-time data. In the sections below, we'll dive into what we believe sets StreamNative apart—from our modern architecture and cloud-native design to our open-source foundation and real-time use cases—and how we see these strengths aligning with Forrester's findings.",[40,82,84],{"id":83},"trusted-by-industry-leaders",[44,85,86],{},"Trusted by Industry Leaders",[48,88,89,90,93,94,97,98,101],{},"Companies across industries are already leveraging StreamNative to drive real-time outcomes. Global enterprises like ",[44,91,92],{},"Cisco"," rely on StreamNative to handle massive IoT telemetry, supporting 245 million+ connected devices. Martech leaders such as ",[44,95,96],{},"Iterable"," process billions of events per day with StreamNative for hyper-personalized customer engagement. And in financial services, ",[44,99,100],{},"FICO"," trusts StreamNative to power its real-time fraud detection and analytics pipelines with a secure, scalable streaming backbone.",[48,103,104,105,108],{},"The Forrester report notes that, “",[36,106,107],{},"Customers appreciate the lower infrastructure costs that result from StreamNative’s cost-efficient, Kafka-compatible architecture. Customers note excellent support responsiveness…","”",[40,110,112],{"id":111},"modern-cloud-native-architecture-built-for-scale",[44,113,114],{},"Modern, Cloud-Native Architecture Built for Scale",[48,116,117],{},"From day one, StreamNative was designed with a modern architecture to meet the demanding scale and flexibility requirements of real-time data. Unlike legacy streaming systems that often rely on tightly coupled storage and compute, StreamNative's platform takes a cloud-native approach: it decouples these layers to enable elastic scalability and efficient resource utilization across any environment. The core is powered by Apache Pulsar—a distributed messaging and streaming engine—enhanced with multi-protocol support (including native Apache Kafka API compatibility) to unify diverse data streams under one roof. This means organizations can consolidate siloed messaging systems and handle both high-volume event streams and traditional message queues on a single platform, without sacrificing performance or reliability.",[48,119,120,121,108],{},"Forrester's evaluation described that “",[36,122,123],{},"StreamNative aims to provide a high-performance, multi-protocol streaming data platform: It uses Apache Pulsar with Kafka API compatibility to deliver cost-efficient, real-time applications for enterprises. It appeals to organizations that want a flexible, low-cost streaming solution, due to its focus on scalability and resource optimization, while its investments in Pulsar’s open-source ecosystem and performance optimization make it the primary platform for enterprises wishing to implement Pulsar.",[48,125,126],{},"Our cloud-first, leaderless architecture (with no single broker bottlenecks) and tiered storage model were built to maximize throughput and cost-efficiency for real-time workloads. By separating compute from storage and leveraging distributed object storage, StreamNative can retain huge volumes of event data indefinitely while keeping compute costs in check—effectively providing a flexible, low-cost streaming solution.",[48,128,129,130,133],{},"This modern design not only delivers high performance, but also ensures fault tolerance and geo-distribution out of the box, so enterprises can trust their streaming data is always available and durable. As Forrester’s evaluation noted, StreamNative ",[36,131,132],{},"\"excels at messaging and resource optimization\" and “Its platform supports use cases like real-time analytics and event-driven architectures with robust scalability.","” Our architecture provides the strong foundation that today's real-time applications demand, from ultra-fast data ingestion to seamless scale-out across hybrid and multi-cloud environments.",[40,135,137],{"id":136},"open-source-foundation-and-pulsar-expertise",[44,138,139],{},"Open Source Foundation and Pulsar Expertise",[48,141,142],{},"StreamNative's DNA is rooted in open source innovation. Our founders are the original creators of Apache Pulsar, and we've built our platform with the same open principles: freedom, flexibility, and community-driven innovation. For developers and data teams, this means adopting StreamNative comes with no proprietary lock-in—instead, you get a platform built on open standards and a thriving ecosystem. We offer broad API compatibility (Pulsar, Kafka, JMS, MQTT, and more) so that teams can work with familiar interfaces and integrate StreamNative into existing systems with ease.",[48,144,145],{},"StreamNative is the primary commercial contributor to the Apache Pulsar project and its surrounding ecosystem. We invest heavily in Pulsar's ongoing improvements our investments in Pulsar's open-source ecosystem and performance optimization bolster StreamNative's value. We also foster a vibrant community through initiatives like the Data Streaming Summit and free training resources.",[48,147,148,149,152,153,156],{},"Forrester's assessment noted that StreamNative’s “",[36,150,151],{},"events-driven agents, extensibility, and performance architecture are solid,","” and we're continuing to build on that foundation. ",[44,154,155],{},"We're actively investing in expanding our tooling for observability, governance, schema management, and developer productivity","—areas we recognize as critical for enterprise adoption and where we're committed to accelerating our roadmap.",[48,158,159,160],{},"Being open also means embracing an open ecosystem of technologies. StreamNative actively integrates with the tools and platforms that matter most to our users. We partner with industry leaders like Snowflake, Databricks, Google, and Ververica to ensure our streaming platform works seamlessly with data warehouses, lakehouse storage, and stream processing frameworks. Forrester’s evaluation observed that StreamNative’s ",[36,161,162],{},"\"investments in Pulsar’s open-source ecosystem and performance optimization make it the primary platform for enterprises wishing to implement Pulsar.\"",[40,164,166],{"id":165},"powering-real-time-use-cases-across-industries",[44,167,168],{},"Powering Real-Time Use Cases Across Industries",[48,170,171],{},"One of the greatest validations of StreamNative's approach is the success our customers are achieving with real-time data. StreamNative's platform is versatile and use-case agnostic—if an application demands high-volume, low-latency data movement, we can power it. This flexibility is why our customer base spans industries from finance and IoT to major automobile manufacturers and online gaming. The common thread is that these organizations need to process and react to data in milliseconds, and StreamNative is delivering the capabilities to make that possible.",[48,173,174],{},"Cisco uses StreamNative to underpin an IoT telemetry system of colossal scale, connecting hundreds of millions of devices and thousands of enterprise clients with real-time data streams. The platform's multi-tenant design and proven reliability allow Cisco to offer its customers a live feed of device data with unwavering confidence. In the financial sector, FICO has built streaming pipelines on StreamNative to detect fraud as transactions happen and to monitor systems in real time. With StreamNative's strong guarantees around message durability and ordering, FICO can catch anomalies or suspicious patterns within seconds. And in digital customer engagement, Iterable relies on StreamNative to process billions of events every day—clicks, views, purchases—so that marketers can trigger personalized campaigns instantly based on user behavior.",[48,176,177],{},"Our customers uniformly deal with mission-critical data streams, where downtime or delays are unacceptable. StreamNative's fault-tolerant, scalable infrastructure has proven equal to the task, handling scenarios like bursting to millions of events per second or seamlessly spanning multiple cloud regions. Forrester's report recognized StreamNative for supporting event-driven architectures with robust scalability—which for us is a reflection of our platform's ability to meet the most demanding enterprise requirements.",[40,179,181],{"id":180},"continuing-to-innovate-ursa-orca-and-the-road-ahead",[44,182,183],{},"Continuing to Innovate: Ursa, Orca, and the Road Ahead",[48,185,186,187,190],{},"While we are thrilled to be recognized in Forrester's Streaming Data Platforms Wave, we view this as just the beginning. StreamNative's vision has always been bold: to ",[44,188,189],{},"provide a unified platform that not only handles today's streaming needs but also anticipates the emerging requirements of tomorrow",".",[48,192,193],{},"One key area of focus is the convergence of streaming data with advanced analytics and AI. As Forrester points out in the report, technology leaders should look for platforms that natively integrate messaging, stream processing, and analytics to provide AI agents with real-time, contextualized information. We couldn't agree more. Our award-winning Ursa Engine and Orca Agent Engine are aimed at extending our platform up the stack—bridging the gap between data streams and data lakes, and between event streams and intelligent processing.",[48,195,196],{},"Our new Ursa Engine introduces a lakehouse-native approach to streaming: it can write events directly to table formats like Iceberg on cloud storage, eliminating entire classes of ETL jobs and making fresh data instantly available for analytics queries. By integrating streaming and lakehouse technologies, we help customers collapse data silos and accelerate their AI\u002FML pipelines.",[48,198,199,200,203],{},"Beyond analytics integration, we are also enhancing StreamNative with more out-of-the-box processing and governance capabilities. In the coming months, we plan to introduce new features for lightweight stream processing and transformation, making it easier to build reactive applications directly on the platform. We're also expanding our ecosystem of connectors and integrations, so that whether your data lands in Snowflake, Databricks, or an AI model, StreamNative will seamlessly feed it. ",[44,201,202],{},"We're investing significantly in enterprise features including security, schema registry, governance, and monitoring tooling","—capabilities that are essential for mission-critical deployments and where we're committed to continued improvement.",[48,205,206],{},"This recognition from Forrester energizes us to keep innovating at full speed. We're sharing this honor with our amazing customers, community, and partners who drive us forward every day. Your feedback and real-world challenges have helped shape StreamNative into what it is today, and together, we will shape the future of streaming data. Thank you for joining us on this journey—we're just getting started, and we can't wait to deliver even more value as we continue to evolve our platform. Onward to real-time everything!",[208,209],"hr",{},[32,211,213],{"id":212},"streamnative-in-the-forrester-wave-evaluation-findings",[44,214,215,216,223],{},"StreamNative in ",[44,217,218],{},[55,219,220],{"href":57},[44,221,222],{},"The Forrester Wave™",": Evaluation Findings",[225,226,228],"h5",{"id":227},"recognized-as-a-contender-among-15-streaming-data-platform-providers","• Recognized as a Contender among 15 streaming data platform providers",[225,230,232],{"id":231},"received-the-highest-scores-possible-50-in-both-the-messaging-and-resource-optimization-criteria","* Received the highest scores possible (5.0) in both the Messaging and Resource Optimization criteria",[225,234,236],{"id":235},"cited-as-the-primary-platform-for-enterprises-wishing-to-implement-pulsar","• Cited as the primary platform for enterprises wishing to implement Pulsar",[225,238,240],{"id":239},"noted-for-excelling-at-messaging-and-resource-optimization","• Noted for excelling at messaging and resource optimization",[225,242,244],{"id":243},"customers-cited-lower-infrastructure-costs-and-excellent-support-responsiveness","• Customers cited lower infrastructure costs and excellent support responsiveness",[225,246,248],{"id":247},"recognized-for-supporting-event-driven-architectures-with-robust-scalability","• Recognized for supporting event-driven architectures with robust scalability",[208,250],{},[252,253,255,256,259,260,190],"h6",{"id":254},"forrester-disclaimer-forrester-does-not-endorse-any-company-product-brand-or-service-included-in-its-research-publications-and-does-not-advise-any-person-to-select-the-products-or-services-of-any-company-or-brand-based-on-the-ratings-included-in-such-publications-information-is-based-on-the-best-available-resources-opinions-reflect-judgment-at-the-time-and-are-subject-to-change-for-more-information-read-about-forresters-objectivity-here","**Forrester Disclaimer: **",[36,257,258],{},"Forrester does not endorse any company, product, brand, or service included in its research publications and does not advise any person to select the products or services of any company or brand based on the ratings included in such publications. Information is based on the best available resources. Opinions reflect judgment at the time and are subject to change",". *For more information, read about Forrester’s objectivity *",[55,261,265],{"href":262,"rel":263},"https:\u002F\u002Fwww.forrester.com\u002Fabout-us\u002Fobjectivity\u002F",[264],"nofollow",[36,266,267],{},"here",[208,269],{},[252,271,273],{"id":272},"apache-apache-pulsar-apache-kafka-apache-flink-and-other-names-are-trademarks-of-the-apache-software-foundation-no-endorsement-by-apache-or-other-third-parties-is-implied",[36,274,275],{},"Apache®, Apache Pulsar®, Apache Kafka®, Apache Flink® and other names are trademarks of The Apache Software Foundation. No endorsement by Apache or other third parties is implied.",{"title":18,"searchDepth":19,"depth":19,"links":277},[278,280,281,282,283,284,285],{"id":34,"depth":279,"text":38},3,{"id":42,"depth":19,"text":46},{"id":83,"depth":19,"text":86},{"id":111,"depth":19,"text":114},{"id":136,"depth":19,"text":139},{"id":165,"depth":19,"text":168},{"id":180,"depth":19,"text":183,"children":286},[287],{"id":212,"depth":279,"text":288},"StreamNative in The Forrester Wave™: Evaluation Findings","Company",null,"2025-12-16","StreamNative is recognized in The Forrester Wave™: Streaming Data Platforms, Q4 2025. Discover why Forrester highlights StreamNative's high-performance messaging, efficient resource use, and cost-effective Kafka API compatibility for real-time innovation.","\u002Fimgs\u002Fblogs\u002F693bd36cf01b217dcb67278f_Streamnative_blog_thumbnail.png",false,{},0,"\u002Fblog\u002Fstreamnative-recognized-in-the-forrester-wave-streaming-data-platforms-2025","10 mins read",{"title":26,"description":292},"blog\u002Fstreamnative-recognized-in-the-forrester-wave-streaming-data-platforms-2025",[302,303,304],"Announcements","Real-Time","Forrester","sOeeJtEO3O-IIfTPJjY1AFOMawZ_rf8FOH8A98NEKgU",{"id":307,"title":308,"authors":309,"body":312,"category":931,"createdAt":290,"date":932,"description":933,"extension":8,"featured":294,"image":934,"isDraft":294,"link":290,"meta":935,"navigation":7,"order":296,"path":936,"readingTime":937,"relatedResources":290,"seo":938,"stem":939,"tags":940,"__hash__":942},"blogs\u002Fblog\u002Fa-comparison-of-transaction-buffer-snapshot-strategies-in-apache-pulsar.md","A Comparison of Transaction Buffer Snapshot Strategies in Apache Pulsar 3.0",[310,311],"Xiangying Meng","Lishen Yao",{"type":15,"value":313,"toc":908},[314,316,330,339,347,350,354,358,361,365,368,372,375,379,383,386,391,394,398,401,405,408,412,415,419,422,425,429,438,441,444,448,521,525,535,539,542,548,551,557,561,579,583,586,589,595,598,601,605,608,612,619,627,631,637,645,649,652,656,659,662,672,680,683,693,701,704,707,711,714,717,723,731,734,740,745,748,751,755,758,761,767,775,778,784,789,792,795,799,802,805,811,819,822,828,833,836,839,843,846,850,853,856,859,863,872],[40,315,46],{"id":42},[48,317,318,323,324,329],{},[55,319,322],{"href":320,"rel":321},"https:\u002F\u002Fpulsar.apache.org\u002Fblog\u002F2023\u002F05\u002F02\u002Fannouncing-apache-pulsar-3-0\u002F",[264],"Pulsar 3.0"," was released on May 2 and introduces a variety of new feature enhancements that improve the performance and stability for teams operating Pulsar at scale as well as making it more stable (and predictable) for powering messaging and data streaming services for mission critical use cases. Among these improvements is  ",[55,325,328],{"href":326,"rel":327},"https:\u002F\u002Fgithub.com\u002Fapache\u002Fpulsar\u002Fissues\u002F16913",[264],"transaction buffer segmented snapshots",". The new design incorporates multiple snapshot segments through a secondary index, with index and snapshot segments stored in different compact topics.",[48,331,332,333,338],{},"The ",[55,334,337],{"href":335,"rel":336},"https:\u002F\u002Fpulsar.apache.org\u002Fdocs\u002F3.0.x\u002Ftransactions\u002F#transaction-buffer",[264],"transaction buffer"," plays an important role in Pulsar transactions, serving as a repository for messages produced within a transaction. In Pulsar releases prior to 3.0, the transaction buffer involves handling messages sent with transactions and taking periodic snapshots to avoid replaying all messages from the original topic. However, when a topic has long-term data retention and many aborted transactions, a single snapshot may become a bottleneck, causing increased costs as the snapshot size grows.",[48,340,341,342,346],{},"To evaluate the effectiveness of using multiple snapshot segments, the engineering team at ",[55,343,345],{"href":344},"\u002F","StreamNative",", who contributed to the release of Pulsar 3.0, performed some benchmark tests using the OpenMessaging Benchmark framework. This benchmark report juxtaposes the new transaction buffer strategy of using multiple snapshots (segmented snapshots) against the previous single snapshot approach, focusing on key performance indicators such as throughput and latency.",[48,348,349],{},"The objective of this report is to offer users insights to select the most suitable strategy for their specific use cases, and to inform decisions regarding future optimizations and enhancements for more efficient transaction buffer management.",[40,351,353],{"id":352},"key-benchmark-findings","Key benchmark findings",[32,355,357],{"id":356},"_75x-improvement-for-network-io-efficiency","7.5x improvement for network IO efficiency",[48,359,360],{},"When tested under the same transaction abort rates, the newly implemented multi-snapshot strategy consistently maintained a steady throughput, averaging at 2 MB\u002Fs, and displayed a regular, periodic oscillation. This is in stark contrast to the previous strategy, which demonstrated an increasing throughput with an average rate of 15 MB\u002Fs. The new strategy, therefore, offers a significant advantage in terms of network IO conservation.",[32,362,364],{"id":363},"_20x-lower-write-latency","20x lower write latency",[48,366,367],{},"The multi-snapshot strategy consistently kept write latency within a narrow band of 10-20ms. This is a marked improvement over the previous strategy, which saw write latency continually growing up to 200ms, an indication of a performance bottleneck.",[32,369,371],{"id":370},"_20x-shorter-garbage-collection-gc-pauses","20x shorter Garbage Collection (GC) pauses",[48,373,374],{},"Throughout the testing period, the GC pauses for the new multi-snapshot strategy consistently hovered around 100ms, demonstrating efficient memory management. By comparison, the previous strategy saw GC pauses that not only increased over time but reached up to 2 seconds and even exceeded 20 seconds after an hour of testing. The consistent and stable performance of the new strategy points to enhanced system stability and operational efficiency.",[40,376,378],{"id":377},"test-overview","Test overview",[32,380,382],{"id":381},"what-we-tested","What we tested",[48,384,385],{},"We chose the following five performance indicators for benchmark testing, as they provide a comprehensive evaluation of system performance and help us better understand how the system behaves under different pressures.",[387,388,390],"h4",{"id":389},"_1-throughput","1. Throughput",[48,392,393],{},"This metric measures the amount of data that a system can handle within a specific timeframe. It is a crucial indicator of a system’s processing power and network efficiency. We anticipated that the new implementation would utilize network IO more efficiently than the existing one (i.e., higher throughput for the single snapshot strategy).",[387,395,397],{"id":396},"_2-entry-size-and-write-latency","2. Entry size and write latency",[48,399,400],{},"Entry size refers to the size of data segments written into the system, while write latency measures the delay between the issue of a write request and the completion of the operation. Smaller entry sizes and lower write latency generally improve system responsiveness and performance. The new implementation was expected to limit the size of snapshot segments, which might reduce write latency compared to the previous one. We expected a reduction in both the entry size and write latency.",[387,402,404],{"id":403},"_3-cpu-usage","3. CPU usage",[48,406,407],{},"This metric quantifies the intensity of Central Processing Unit (CPU) utilization by the system. It’s a critical metric as both the new and old strategies can potentially impact CPU usage. Under the correct configurations, we anticipated that the CPU usage of the new strategy would not exceed that of the old one and would be more stable.",[387,409,411],{"id":410},"_4-gc-pauses","4. GC pauses",[48,413,414],{},"Garbage Collection (GC) is an automatic memory management method to free up memory no longer in use or needed. GC pauses occur when GC operations pause the program to perform memory clean-up, which could negatively impact system performance. By monitoring these pauses, we can understand how the system manages memory and maintains performance. We expected a decrease in GC pauses with the new implementation.",[387,416,418],{"id":417},"_5-heap-memory","5. Heap memory",[48,420,421],{},"Heap memory refers to the runtime data area from which memory for all class instances and arrays is allocated. High heap usage could signal memory leaks, inadequate sizing, or code that creates excessive temporary objects. Therefore, tracking heap memory usage is crucial to ensuring effective use of memory resources. We expected a decrease in heap memory usage.",[48,423,424],{},"In summary, we hoped that these tests could demonstrate the superior performance of the new strategy, including lower throughput, reduced write latency, optimized CPU usage, fewer GC pauses, and lower heap memory usage.",[32,426,428],{"id":427},"how-we-set-up-the-tests","How we set up the tests",[48,430,431,432,437],{},"We conducted all tests using the ",[55,433,436],{"href":434,"rel":435},"https:\u002F\u002Fgithub.com\u002Fopenmessaging\u002Fbenchmark",[264],"OpenMessaging Benchmark framework"," with the hosted service on StreamNative Cloud. The test environments for both snapshot strategies were identical in terms of infrastructure configurations and benchmark settings. In each case, we used a hosted Pulsar cluster deployed on Kubernetes, comprising 3 broker Pods, 3 bookie Pods, and 3 ZooKeeper Pods. We used Grafana to provide observability for the necessary metrics.",[48,439,440],{},"Given that snapshots are used to store information of aborted transactions, which will be cleared when the original transactional message ledger is deleted, we tested snapshot strategies under conditions of high transaction abort frequency and long retention time.",[48,442,443],{},"See the following configurations for the benchmark testbed details.",[387,445,447],{"id":446},"infrastructure","Infrastructure",[449,450,451,455,458,461,464,467,470,473,476,479,482,485,488,491,494,496,499,501,504,507,510,513,515,518],"ul",{},[452,453,454],"li",{},"StreamNative Cloud: Hosted service",[452,456,457],{},"~Infrastructure vendor: Google Cloud",[452,459,460],{},"~Settings: Advanced with Transactions enabled",[452,462,463],{},"Image: streamnative\u002Fpulsar-cloud:3.0.0.3-SNAPSHOT",[452,465,466],{},"Kubernetes version: 1.24.11-gke.1000",[452,468,469],{},"Network speed: 30 Gbps",[452,471,472],{},"Pulsar cluster components:",[452,474,475],{},"~3 broker Pods, each with:",[452,477,478],{},"~~CPU request: 4 cores",[452,480,481],{},"~~Memory request: 4Gi",[452,483,484],{},"~~Heap size: 2G",[452,486,487],{},"~~Direct memory size: 2G",[452,489,490],{},"~3 bookie Pods, each with:",[452,492,493],{},"~~CPU request: 2 cores",[452,495,481],{},[452,497,498],{},"~~Heap size: 1G",[452,500,487],{},[452,502,503],{},"~~1 volume for the journal (default size: 128Gi) and 1 volume for the ledger (default size 1Ti)",[452,505,506],{},"~3 ZooKeeper Pods, each with:",[452,508,509],{},"~~CPU request: 500m",[452,511,512],{},"~~Memory request: 1Gi",[452,514,498],{},[452,516,517],{},"~~Direct memory size: 1G",[452,519,520],{},"Observability tool: Grafana",[387,522,524],{"id":523},"benchmark-settings","Benchmark settings",[526,527,532],"pre",{"className":528,"code":530,"language":531},[529],"language-text","consumerPerSubscription: 0\nmessageSize: 100B\npartitionsPerTopic: 3\npath: workloads\u002Ftransaction-3-topic-3-partitions-100b.yaml\npayloadFile: payload\u002Fpayload-100b.data\nproducerRate: 3000\nproducersPerTopic: 1\nsubscriptionsPerTopic: 0\ntestDurationMinutes: 6000\ntopics: 3\n","text",[533,534,530],"code",{"__ignoreMap":18},[387,536,538],{"id":537},"transaction-buffer-snapshot-settings","Transaction buffer snapshot settings",[48,540,541],{},"Single snapshot strategy config:",[526,543,546],{"className":544,"code":545,"language":531},[529],"PULSAR_PREFIX_transactionBufferSegmentedSnapshotEnabled: \"false\"\nPULSAR_PREFIX_transactionCoordinatorEnabled: \"true\"\nPULSAR_PREFIX_transactionBufferSnapshotSegmentSize: \"51200\"\nPULSAR_PREFIX_maxMessageSize: \"52428800\"\nPULSAR_PREFIX_maxMessagePublishBufferSizeInMB: \"52428800\"\nPULSAR_PREFIX_numIOThreads: \"8\"\nPULSAR_PREFIX_transactionBufferSnapshotMinTimeInMillis=5000\nPULSAR_PREFIX_transactionBufferSnapshotMaxTransactionCount=1000\nbookkeeper.PULSAR_PREFIX_nettyMaxFrameSizeBytes: \"52531200\"\n",[533,547,545],{"__ignoreMap":18},[48,549,550],{},"Multi-snapshot strategy config:",[526,552,555],{"className":553,"code":554,"language":531},[529],"PULSAR_PREFIX_transactionBufferSegmentedSnapshotEnabled: \"true\"\nPULSAR_PREFIX_transactionCoordinatorEnabled: \"true\"\nPULSAR_PREFIX_transactionBufferSnapshotSegmentSize: \"1024000\"\nPULSAR_PREFIX_numIOThreads: \"8\"\nPULSAR_PREFIX_transactionBufferSnapshotMinTimeInMillis=5000\nPULSAR_PREFIX_transactionBufferSnapshotMaxTransactionCount=1000\n",[533,556,554],{"__ignoreMap":18},[32,558,560],{"id":559},"test-procedures","Test procedures",[562,563,564,567,570,573,576],"ol",{},[452,565,566],{},"Set up the test environment with the specified hardware, software, and network configurations.",[452,568,569],{},"Configured the Pulsar Benchmark tool with the selected parameter settings.",[452,571,572],{},"Conducted performance tests for each scenario and compared the multi-snapshot strategy with the single snapshot strategy.",[452,574,575],{},"Monitored and recorded metrics such as throughput, write latency, and entry size.",[452,577,578],{},"Analyzed the results and drew conclusions according to the performance of both strategies.",[40,580,582],{"id":581},"benchmark-tests-and-results","Benchmark tests and results",[48,584,585],{},"We ran the following benchmark tests with both transaction buffer snapshot strategies.",[48,587,588],{},"The test using the previous strategy ran stably for 70 minutes with the same message send rate, transaction abort rate, and snapshot take rate, followed by an unstable hour (see the CPU usage section below). According to the logs, BookKeeper kept reconnecting during this hour:",[526,590,593],{"className":591,"code":592,"language":531},[529],"io.netty.channel.unix.Errors$NativeIoException: recvAddress(..) failed: Connection reset by peer\n",[533,594,592],{"__ignoreMap":18},[48,596,597],{},"The new multi-snapshot solution ran stably for 160 minutes until the maximum heap memory size was reached.",[48,599,600],{},"We compared the test results from the following five aspects: throughput, latency, CPU usage, GC pauses, and memory.",[32,602,604],{"id":603},"throughput","Throughput",[48,606,607],{},"This benchmark test compared the throughput of both strategies in writing snapshots to the system topic, with the same message send rates and transaction abort rates. We anticipated that the segmented snapshots would have significantly lower throughput, resulting in substantial network IO cost savings.",[387,609,611],{"id":610},"single-snapshot-strategy","Single snapshot strategy",[48,613,614,618],{},[615,616],"img",{"alt":18,"src":617},"\u002Fimgs\u002Fblogs\u002F6482835369662b852dcdec0d_image11.webp","Figure 1. Single snapshot strategy - Publish rate and throughput",[449,620,621,624],{},[452,622,623],{},"The publish rate gradually declined from 3.5msg\u002Fs to 2msg\u002Fs.",[452,625,626],{},"The publish throughput increased linearly, reaching up to 25 MB\u002Fs.",[387,628,630],{"id":629},"multi-snapshot-strategy","Multi-snapshot strategy",[48,632,633,636],{},[615,634],{"alt":18,"src":635},"\u002Fimgs\u002Fblogs\u002F648283b4ff4192b3250991a8_image5.webp","Figure 2. Multi-snapshot strategy - Publish rate and throughput",[449,638,639,642],{},[452,640,641],{},"The publish rate remained stable at approximately 3.5msg\u002Fs.",[452,643,644],{},"The publish throughput saw a periodic change from 0MB\u002Fs to 4MB\u002Fs.",[387,646,648],{"id":647},"analysis","Analysis",[48,650,651],{},"After one hour of testing, the multi-snapshot strategy showed a throughput that was an order of magnitude lower than the single snapshot strategy, and it demonstrated greater stability. The throughput of the single snapshot strategy eventually increased to 25MB\u002Fs, but with a decreasing message publish rate. In contrast, the multi-snapshot strategy's throughput periodically fluctuated within the range of 5MB\u002Fs, as per the configuration, while maintaining a stable message publish rate. This implies that the new strategy can conserve more network IO resources used for sending messages to the system topic, resulting in improved and more stable performance.",[32,653,655],{"id":654},"entry-size-and-write-latency","Entry size and write latency",[48,657,658],{},"This test focused on snapshot entry size and write latency. We set the snapshot segment size to 1MB (1,024,000 bytes) and did not impose any restrictions on the entry size for single snapshot, to avoid test interruption due to errors caused by excessively large entries. Our expectation was that the new segmented snapshot solution would maintain stability in snapshot segment size, thus ensuring consistently low latency.",[387,660,611],{"id":661},"single-snapshot-strategy-1",[48,663,664,667,668,671],{},[615,665],{"alt":18,"src":666},"\u002Fimgs\u002Fblogs\u002F648284086cf53081242ecb8d_image3.png","Figure 3. Single snapshot strategy - Write latency",[615,669],{"alt":18,"src":670},"\u002Fimgs\u002Fblogs\u002F648284306d987582a9d57be2_image2.png","Figure 4. Single snapshot strategy - Entry size",[449,673,674,677],{},[452,675,676],{},"With the previous single snapshot strategy, the storage writes latency increased over time. After running for an hour, just before the benchmark crashed, the write latency was mostly between 100ms and 200ms.",[452,678,679],{},"The size of the snapshot entry in the previous strategy also increased as the test progressed. After 10 minutes, the size exceeded the observable maximum value of 1MB.",[387,681,630],{"id":682},"multi-snapshot-strategy-1",[48,684,685,688,689,692],{},[615,686],{"alt":18,"src":687},"\u002Fimgs\u002Fblogs\u002F64828452452fc5df9fca521b_image4.png","Figure 5. Multi-snapshot strategy - Write latency",[615,690],{"alt":18,"src":691},"\u002Fimgs\u002Fblogs\u002F6482847007b33fd2344bfdbb_image9.webp","Figure 6. Multi-snapshot strategy - Entry size",[449,694,695,698],{},[452,696,697],{},"The latency of the new approach showed periodic fluctuations but never exceeded the range of 10-20ms.",[452,699,700],{},"As we set the transactionBufferSnapshotSegmentSize to 1024000, the entry size was always less than 1MB.",[387,702,648],{"id":703},"analysis-1",[48,705,706],{},"The test results indicate that the snapshot entry size in the new strategy did not continuously increase like in the single snapshot strategy. It consistently maintained a very low and stable latency. On the other hand, the single snapshot strategy experienced increasing entry size and write latency over time.",[32,708,710],{"id":709},"cpu-usage","CPU usage",[48,712,713],{},"This test compared the CPU utilization between the two snapshot strategy. The number of keys during compaction and the size of sent entries both impact CPU utilization. The multi-snapshot strategy stores data in multiple entries, resulting in multiple keys. By contrast, the single snapshot strategy stores all data in a single entry, leading to write amplification. We expected that, under reasonable configurations, such as a snapshot segment size of 1MB, the new strategy would exhibit lower and more stable CPU utilization.",[387,715,611],{"id":716},"single-snapshot-strategy-2",[48,718,719,722],{},[615,720],{"alt":18,"src":721},"\u002Fimgs\u002Fblogs\u002F648284a89ed007962ff3d115_image10.webp","Figure 7. Single snapshot strategy - CPU usage",[449,724,725,728],{},[452,726,727],{},"During the first 50 minutes of the test, the CPU usage stayed around 200%, with a slight increase.",[452,729,730],{},"After an hour, the CPU usage showed significant fluctuations, and the broker began to become unstable.",[387,732,630],{"id":733},"multi-snapshot-strategy-2",[48,735,736,739],{},[615,737],{"alt":18,"src":738},"\u002Fimgs\u002Fblogs\u002F648284db402ad24821430488_image1.webp","Figure 8. Multi-snapshot strategy - CPU usage",[449,741,742],{},[452,743,744],{},"The CPU usage was stable at about 200% and the test ended due to heap memory OOM.",[387,746,648],{"id":747},"analysis-2",[48,749,750],{},"The test results indicate that, under normal circumstances, the CPU utilization of the new strategy was slightly lower than that of the previous strategy, and it remained more stable before reaching machine bottlenecks. This means that the new multi-snapshot strategy outperforms the previous single snapshot strategy in terms of CPU usage.",[32,752,754],{"id":753},"gc-pauses","GC pauses",[48,756,757],{},"This test examined the GC pause behavior of both strategies. The size and quantity of temporary objects generated during the snapshot-taking process can impact GC pauses. The test was conducted with a snapshot segment size of 1MB, without any restrictions imposed on the size of messages.",[387,759,611],{"id":760},"single-snapshot-strategy-3",[48,762,763,766],{},[615,764],{"alt":18,"src":765},"\u002Fimgs\u002Fblogs\u002F6482850d452fc5df9fcb0bfb_image8.webp","Figure 9. Single snapshot strategy - GC pauses",[449,768,769,772],{},[452,770,771],{},"During the stable testing period, GC pauses kept increasing, peaking at around 2 seconds.",[452,773,774],{},"After an hour of testing, the test became unstable, and the maximum GC pauses reached approximately 20 seconds.",[387,776,630],{"id":777},"multi-snapshot-strategy-3",[48,779,780,783],{},[615,781],{"alt":18,"src":782},"\u002Fimgs\u002Fblogs\u002F64828530fbcf75f737c6f77c_image7.webp","Figure 10. Multi-snapshot strategy - GC pauses",[449,785,786],{},[452,787,788],{},"The delay of the new strategy consistently maintained below 100ms throughput the test.",[387,790,648],{"id":791},"analysis-3",[48,793,794],{},"The test results reveal that the GC pause of the new strategy consistently hovered around 100ms, indicating stable performance. In contrast, the GC pause of the previous single snapshot strategy progressively increased. This is probably caused by the write amplification issue, where the size of temporary snapshot objects generated during each operation continually expanded. Furthermore, as the machine approached its performance bottleneck, GC frequency increased significantly, leading to longer GC pauses. This clearly illustrates the superior performance of the new implementation in terms of managing GC pauses.",[32,796,798],{"id":797},"heap-memory","Heap memory",[48,800,801],{},"This test compared the heap memory growth between the two strategies. The size and quantity of temporary objects generated during the snapshot-taking process can impact heap memory consumption. The test was conducted with a snapshot segment size of 1MB, without any restrictions imposed on the size of messages.",[387,803,611],{"id":804},"single-snapshot-strategy-4",[48,806,807,810],{},[615,808],{"alt":18,"src":809},"\u002Fimgs\u002Fblogs\u002F64828568452fc5df9fcb7394_image12.webp","Figure 11. Single snapshot strategy - Heap memory",[449,812,813,816],{},[452,814,815],{},"After 15 minutes of testing, the heap memory reached 1.5 GB.",[452,817,818],{},"Approximately 6 hours and 50 minutes into the test, the heap memory reached the OOM threshold of 2 GB. It was at this point that the CPU usage and GC pauses began to increase sharply and continued to fluctuate significantly.",[387,820,630],{"id":821},"multi-snapshot-strategy-4",[48,823,824,827],{},[615,825],{"alt":18,"src":826},"\u002Fimgs\u002Fblogs\u002F64828598a1ce59e9ea87d5e1_image6.webp","Figure 12. Multi-snapshot strategy - Heap memory",[449,829,830],{},[452,831,832],{},"The memory of the new strategy steadily increased until the end of the test after OOM.",[387,834,648],{"id":835},"analysis-4",[48,837,838],{},"The test results show that the new strategy exhibits a slower and more stable heap memory growth pattern. By contrast, the previous strategy experienced a faster and more fluctuating heap memory growth. This implies that the new strategy offers advantages in terms of slower and more controlled memory growth, thus ensuring stability and delaying the onset of performance bottlenecks.",[40,840,842],{"id":841},"future-improvement-on-demand-snapshot-segment-loading","Future improvement: On-demand snapshot segment loading",[48,844,845],{},"To reduce startup time when loading transaction buffers, we recommend on-demand snapshot segment loading. The current implementation may read all snapshot segments at startup, leading to longer startup time. With on-demand loading, we can selectively read specific snapshot segments as required, thereby reducing startup time.",[40,847,849],{"id":848},"conclusion","Conclusion",[48,851,852],{},"Our test results demonstrate that the newly implemented multi-snapshot approach significantly outperforms the previous single snapshot approach in key performance metrics. With the same snapshot frequency, the new solution resolves the write amplification issue, resulting in lower network bandwidth utilization, reduced message latency, shorter GC stop-the-world (STW) times, and a more stable memory growth that avoids frequent garbage collection.",[48,854,855],{},"Further optimization, such as the implementation of on-demand loading of snapshot segments and distributed caching, can enhance the performance and stability of the new strategy in transaction buffers.",[48,857,858],{},"In real-world applications, we strongly recommend the adoption of the new multi-snapshot strategy.",[40,860,862],{"id":861},"more-resources","More resources",[48,864,865,866,871],{},"Pulsar has become ",[55,867,870],{"href":868,"rel":869},"https:\u002F\u002Fblogs.apache.org\u002Ffoundation\u002Fentry\u002Fapache-in-2021-by-the",[264],"one of the most active Apache projects"," over the past few years, with a vibrant community driving innovation and improvements to the project. Check out the following resources to learn more about Pulsar.",[449,873,874,881,889,900],{},[452,875,876,877,190],{},"Run fully managed Pulsar services and enable transactions with ",[55,878,880],{"href":879},"\u002Fproduct","StreamNative Cloud",[452,882,883,884,190],{},"Start your on-demand Pulsar training today with ",[55,885,888],{"href":886,"rel":887},"https:\u002F\u002Fwww.academy.streamnative.io\u002F",[264],"StreamNative Academy",[452,890,891,895,896],{},[892,893,894],"span",{},"Blog"," ",[55,897,899],{"href":898},"\u002Fblog\u002Fdeep-dive-into-transaction-buffer-apache-pulsar","A Deep Dive into Transaction Buffer in Apache Pulsar",[452,901,902,895,904],{},[892,903,894],{},[55,905,907],{"href":906},"\u002Fblog\u002Fdeep-dive-transaction-coordinators-apache-pulsar","A Deep Dive into Transaction Coordinators in Apache Pulsar",{"title":18,"searchDepth":19,"depth":19,"links":909},[910,911,916,921,928,929,930],{"id":42,"depth":19,"text":46},{"id":352,"depth":19,"text":353,"children":912},[913,914,915],{"id":356,"depth":279,"text":357},{"id":363,"depth":279,"text":364},{"id":370,"depth":279,"text":371},{"id":377,"depth":19,"text":378,"children":917},[918,919,920],{"id":381,"depth":279,"text":382},{"id":427,"depth":279,"text":428},{"id":559,"depth":279,"text":560},{"id":581,"depth":19,"text":582,"children":922},[923,924,925,926,927],{"id":603,"depth":279,"text":604},{"id":654,"depth":279,"text":655},{"id":709,"depth":279,"text":710},{"id":753,"depth":279,"text":754},{"id":797,"depth":279,"text":798},{"id":841,"depth":19,"text":842},{"id":848,"depth":19,"text":849},{"id":861,"depth":19,"text":862},"Apache Pulsar","2023-06-09","This benchmark report provides an in-depth comparison of the single snapshot strategy and the segmented snapshot strategy in the transaction buffer.","\u002Fimgs\u002Fblogs\u002F649a9195b97b0a0b29c49d2e_a-comparison-of-transaction-buffer-snapshot-strategies-in-apache-pulsar.jpg",{},"\u002Fblog\u002Fa-comparison-of-transaction-buffer-snapshot-strategies-in-apache-pulsar","10 min read",{"title":308,"description":933},"blog\u002Fa-comparison-of-transaction-buffer-snapshot-strategies-in-apache-pulsar",[931,941],"Transactions","fNgPFh8lGbPzWEhFCw7Lxc803LmiGBqQZ10BhOwJTds",{"id":944,"title":945,"authors":946,"body":951,"category":880,"createdAt":290,"date":1701,"description":1702,"extension":8,"featured":7,"image":1703,"isDraft":294,"link":290,"meta":1704,"navigation":7,"order":296,"path":1705,"readingTime":1706,"relatedResources":290,"seo":1707,"stem":1708,"tags":1709,"__hash__":1712},"blogs\u002Fblog\u002Fhow-we-run-a-5-gb-s-kafka-workload-for-just-50-per-hour.md","How We Run a 5 GB\u002Fs Kafka Workload for Just $50 per Hour",[947,948,949,950],"Matteo Meril","Neng Lu","Hang Chen","Penghui Li",{"type":15,"value":952,"toc":1671},[953,956,959,962,965,968,971,974,982,987,990,998,1003,1007,1014,1017,1020,1028,1032,1035,1040,1044,1047,1050,1053,1056,1065,1069,1072,1083,1086,1090,1093,1096,1107,1110,1114,1118,1126,1129,1133,1141,1170,1174,1177,1182,1186,1189,1193,1196,1199,1204,1213,1218,1221,1224,1235,1239,1242,1253,1257,1260,1263,1268,1271,1300,1304,1306,1312,1315,1320,1325,1328,1332,1346,1350,1361,1365,1380,1389,1400,1403,1406,1410,1413,1416,1427,1430,1433,1436,1441,1446,1450,1453,1470,1474,1488,1493,1497,1508,1511,1527,1531,1542,1547,1552,1560,1562,1565,1569,1576,1580,1583,1592,1597,1604,1610,1619,1628,1637,1646,1655,1663],[48,954,955],{},"The rise of DeepSeek has shaken the AI infrastructure market, forcing companies to confront the escalating costs of training and deploying AI models. But the real pressure point isn’t just compute—it’s data acquisition and ingestion costs.",[48,957,958],{},"As businesses rethink their AI cost-containment strategies, real-time data streaming is emerging as a critical enabler. The growing adoption of Kafka as a standard protocol has expanded cost-efficient options, allowing companies to optimize streaming analytics while keeping expenses in check.",[48,960,961],{},"Ursa, the data streaming engine powering StreamNative’s managed Kafka service, is built for this new reality. With its leaderless architecture and native lakehouse storage integration, Ursa eliminates costly inter-zone network traffic for data replication and client-to-broker communication while ensuring high availability at minimal operational cost.",[48,963,964],{},"In this blog post, we benchmarked the infrastructure cost and total cost of ownership (TCO) for running a 5GB\u002Fs Kafka workload across different Kafka vendors, including Redpanda, Confluent WarpStream, and AWS MSK. Our benchmark results show that Ursa can sustain 5GB\u002Fs Kafka workloads at just 5% of the cost of traditional streaming engines like Redpanda—making it the ideal solution for high-performance, cost-efficient ingestion and data streaming for data lakehouses and AI workloads.",[48,966,967],{},"Note: We also evaluated vanilla Kafka in our benchmark; however, for simplicity, we have focused our cost comparison on vendor solutions rather than self-managed deployments. That said, it is important to highlight that both Redpanda and vanilla Kafka use a leader-based data replication approach. In a data-intensive, network-bound workload like 5GB\u002Fs streaming, with the same machine type and replication factor, Redpanda and vanilla Kafka produced nearly identical cost profiles.",[40,969,970],{"id":352},"Key Benchmark Findings",[48,972,973],{},"Ursa delivered 5 GB\u002Fs of sustained throughput at an infrastructure cost of just $54 per hour. For comparison:",[449,975,976,979],{},[452,977,978],{},"MSK: $303 per hour → 5.6x more expensive compared to Ursa",[452,980,981],{},"Redpanda: $988 per hour → 18x more expensive compared to Ursa",[48,983,984],{},[615,985],{"alt":18,"src":986},"\u002Fimgs\u002Fblogs\u002F679c71b67d9046f26edc7977_AD_4nXfvTqyBNUBu2lObdkKAx-5UNkpNP8UYULLZyOcixE6z99VMZUUEsUqWjzexI7vjyNGRNSAUoM9smYvdTP55ctAhIbrs5lmQgcSVMWdaoigbWouCl95DVSQsxooY-qqfGcYqS4g4zA.png",[48,988,989],{},"Beyond infrastructure costs, when factoring in both storage pricing, vendor pricing and operational expenses, Ursa’s total cost of ownership (TCO) for a 5GB\u002Fs workload with a 7-day retention period is:",[449,991,992,995],{},[452,993,994],{},"50% cheaper than Confluent WarpStream",[452,996,997],{},"85% cheaper than MSK and Redpanda",[48,999,1000],{},[615,1001],{"alt":18,"src":1002},"\u002Fimgs\u002Fblogs\u002F679c602d77e9c706de5343b8_AD_4nXeDv8rrv_C1CTCCiqYo1zpvlGYbdBk1r0VEqovAPu22iFMQZgh54Hfw9PBMLzM7jDFxKwAFDxbdG0np4XVk_tGsWhEKMloLRcmmea7lvueCx-0cFsyaE3Mya4Mxc1Dox95A6JEc.png",[40,1004,1006],{"id":1005},"ursa-highly-cost-efficient-data-streaming-at-scale","Ursa: Highly Cost-Efficient Data Streaming at Scale",[48,1008,1009,1013],{},[55,1010,1012],{"href":1011},"\u002Fblog\u002Fursa-reimagine-apache-kafka-for-the-cost-conscious-data-streaming","Ursa"," is a next-generation data streaming engine designed to deliver high performance at a fraction of the cost of traditional disk-based solutions. It is fully compatible with Apache Kafka and Apache Pulsar APIs, while leveraging a leaderless, lakehouse-native architecture to maximize scalability, efficiency, and cost savings.",[48,1015,1016],{},"Ursa’s key innovation is separating storage from compute and decoupling metadata\u002Findex operations from data operations by utilizing cloud object storage (e.g., AWS S3) instead of costly inter-zone disk-based replication. It also employs open lakehouse formats (Iceberg and Delta Lake), enabling columnar compression to significantly reduce storage costs while maintaining durability and availability.",[48,1018,1019],{},"In contrast, traditional streaming systems—like Kafka and Redpanda—depend on leader-based architectures, which drive up inter-zone traffic costs due to replication and client communication. Ursa mitigates these costs by:",[449,1021,1022,1025],{},[452,1023,1024],{},"Eliminating inter-zone traffic costs via a leaderless architecture.",[452,1026,1027],{},"Replacing costly inter-zone replication with direct writes to cloud storage using open lakehouse formats.",[40,1029,1031],{"id":1030},"how-ursa-eliminates-inter-zone-traffic","How Ursa Eliminates Inter-Zone Traffic",[48,1033,1034],{},"Ursa minimizes inter-zone traffic by leveraging a leaderless architecture, which eliminates inter-zone communication between clients and brokers, and lakehouse-native storage, which removes the need for inter-zone data replication. This approach ensures high availability and scalability while avoiding unnecessary cross-zone data movement.",[48,1036,1037],{},[615,1038],{"alt":18,"src":1039},"\u002Fimgs\u002Fblogs\u002F679c602e21b3571bb7117dca_AD_4nXd7Oahc77NjRLNvA9clLt0tsyU6MrIqVibFYv5pW5giTIcCHPr3EA_yTGzfVEUIVO3VXK56qWK8zmBCp5lY0E_4nmlWIPFrHjtHylA5NhwELjn-UB0fLG2h_kbrxrc7Cs_edvveNA.png",[32,1041,1043],{"id":1042},"leaderless-architecture","Leaderless architecture",[48,1045,1046],{},"Traditional streaming engines such as Kafka, Pulsar, or RedPanda rely on a leader-based model, where each partition is assigned to a single leader broker that handles all writes and reads.",[48,1048,1049],{},"Pros of Leader-Based Architectures:\n✔ Maintains message ordering via local sequence IDs\n✔ Delivers low latency and high performance through message caching",[48,1051,1052],{},"Cons of Leader-Based Architectures:\n✖ Throughput bottlenecked by a single broker per partition\n✖ Inter-zone traffic required for high availability in multi-AZ deployments",[48,1054,1055],{},"While Kafka and Pulsar offer partial solutions (e.g., reading from followers, shadow topics) to reduce read-related inter-zone traffic, producers still send data to a single leader.",[48,1057,1058,1059,1064],{},"Ursa removes the concept of topic ownership, allowing any broker in the cluster to handle reads or writes for any partition. The primary challenge—ensuring message ordering—is solved with ",[55,1060,1063],{"href":1061,"rel":1062},"https:\u002F\u002Fgithub.com\u002Fstreamnative\u002Foxia",[264],"Oxia",", a scalable metadata and index service created by StreamNative in 2022.",[32,1066,1068],{"id":1067},"oxia-the-metadata-layer-enabling-leaderless-architecture","Oxia: The Metadata Layer Enabling Leaderless Architecture",[48,1070,1071],{},"Ensuring message ordering in a leaderless architecture is complex, but Ursa solves this with Oxia:",[449,1073,1074,1077,1080],{},[452,1075,1076],{},"Handles millions of metadata\u002Findex operations per second",[452,1078,1079],{},"Generates sequential IDs to maintain strict message ordering",[452,1081,1082],{},"Optimized for Kubernetes with horizontal scalability",[48,1084,1085],{},"Producers and consumers can connect to any broker within their local AZ, eliminating inter-zone traffic costs while maintaining performance through localized caching.",[32,1087,1089],{"id":1088},"zero-interzone-data-replication","Zero interzone data replication",[48,1091,1092],{},"In most distributed systems, data replication from a leader (primary) to followers (replicas) is crucial for fault tolerance and availability. However, replication across zones can inflate infrastructure expenses substantially.",[48,1094,1095],{},"Ursa avoids these costs by writing data directly to cloud storage (e.g., AWS S3, Google GCS):",[449,1097,1098,1101,1104],{},[452,1099,1100],{},"Built-In Resilience: Cloud storage inherently offers high availability and fault tolerance without inter-zone traffic fees.",[452,1102,1103],{},"Tradeoff: Slightly higher latency (sub-second, with p99 at 500 milliseconds) compared to local disk\u002FEBS (single-digit to sub-100 milliseconds), in exchange for significantly lower costs (up to 10x lower).",[452,1105,1106],{},"Flexible Modes: Ursa is an addition to the classic BookKeeper-based engine, providing users with the flexibility to optimize for either cost or low latency based on their workload requirements.",[48,1108,1109],{},"By foregoing conventional replication, Ursa slashes inter-zone traffic costs and associated complexities—making it a compelling option for organizations seeking to balance high-performance data streaming with strict budget constraints.",[40,1111,1113],{"id":1112},"how-we-ran-a-5-gbs-test-with-ursa","How We Ran a 5 GB\u002Fs Test with Ursa",[32,1115,1117],{"id":1116},"ursa-cluster-deployment","Ursa Cluster Deployment",[449,1119,1120,1123],{},[452,1121,1122],{},"9 brokers across 3 availability zones, each on m6i.8xlarge (Fixed 12.5 Gbps bandwidth, 32 vCPU cores, 128 GB memory).",[452,1124,1125],{},"Oxia cluster (metadata store) with 3 nodes of m6i.8xlarge, distributed across three availability zones (AZs).",[48,1127,1128],{},"During peak throughput (5 GB\u002Fs), each broker’s network usage was about 10 Gbps.",[32,1130,1132],{"id":1131},"openmessaging-benchmark-workers-configuration","OpenMessaging Benchmark Workers & Configuration",[48,1134,1135,1136,1140],{},"The OpenMessaging Benchmark(OMB) Framework is a suite of tools that make it easy to benchmark distributed messaging systems in the cloud. Please check ",[55,1137,1138],{"href":1138,"rel":1139},"https:\u002F\u002Fopenmessaging.cloud\u002Fdocs\u002Fbenchmarks\u002F",[264]," for details.",[449,1142,1143,1158,1167],{},[452,1144,1145,1146,1151,1152,1157],{},"12 OMB workers: 6 for ",[55,1147,1150],{"href":1148,"rel":1149},"https:\u002F\u002Fgist.github.com\u002Fcodelipenghui\u002Fd1094122270775e4f1580947f80c5055",[264],"producers",", 6 for ",[55,1153,1156],{"href":1154,"rel":1155},"https:\u002F\u002Fgist.github.com\u002Fcodelipenghui\u002F06bada89381fb77a7862e1b4c1d8963d",[264],"consumers"," across 3 availability zones, on m6i.8xlarge instances. Each worker is configured with 12 CPU cores and 48 GB memory.",[452,1159,1160,1161,1166],{},"Sample YAML ",[55,1162,1165],{"href":1163,"rel":1164},"https:\u002F\u002Fgist.github.com\u002Fcodelipenghui\u002F204c1f26c4d44a218ae235bf2de99904",[264],"scripts"," provided for Kafka-compatible configuration and rate limits.",[452,1168,1169],{},"Achieved consistent 5 GB\u002Fs publish\u002Fsubscribe throughput.",[40,1171,1173],{"id":1172},"ursa-benchmark-tests-results","Ursa Benchmark Tests & Results",[48,1175,1176],{},"The following diagram demonstrates that Ursa can consistently handle 5 GB\u002Fs of traffic, fully saturating the network across all broker nodes.",[48,1178,1179],{},[615,1180],{"alt":18,"src":1181},"\u002Fimgs\u002Fblogs\u002F679c602d7b261bac1113f7d6_AD_4nXdDPsRc3koXICiFF0bqSmGWbJt_RlUy4FE3ruuWOfbCfpcqZ1dejjqGbkaCJv2hQFL1nirRouBVRW2l5uMWBvY9naMqGB_wHcLI14dBM0f85TXhmdm3UxEv1yGX9Y4hf5FttSkZew.png",[40,1183,1185],{"id":1184},"comparing-infrastructure-cost","Comparing Infrastructure Cost",[48,1187,1188],{},"This benchmark first evaluates infrastructure costs of running a 5 GB\u002Fs streaming workload (1:1 producer-to-consumer ratio) across different data streaming engines, including Ursa, Redpanda, and AWS MSK, with a focus on multi-AZ deployments to ensure a fair comparison.",[32,1190,1192],{"id":1191},"test-setup-key-assumptions","Test Setup & Key Assumptions",[48,1194,1195],{},"All tests use multi-AZ configurations, with clusters and clients distributed across three AWS availability zones (AZs). Cluster size scales proportionally to the number of AZs, and rack-awareness is enabled for all engines to evenly distribute topic partitions and leaders.",[48,1197,1198],{},"To ensure a fair comparison, we selected the same machine type capable of fully utilizing both network and storage bandwidth for Ursa and Redpanda in this 5GB\u002Fs test:",[449,1200,1201],{},[452,1202,1203],{},"9 × m6i.8xlarge instances",[48,1205,1206,1207,1212],{},"However, MSK's storage bandwidth limits vary depending on the selected instance type, with the highest allowed limit capped at 1000 MiB\u002Fs per broker, according to",[55,1208,1211],{"href":1209,"rel":1210},"https:\u002F\u002Fdocs.aws.amazon.com\u002Fmsk\u002Flatest\u002Fdeveloperguide\u002Fmsk-provision-throughput-management.html#throughput-bottlenecks",[264]," AWS documentation",". Given this constraint, achieving 5 GB\u002Fs throughput with a replication factor of 3 required the following setup:",[449,1214,1215],{},[452,1216,1217],{},"15 × kafka.m7g.8xlarge (32 vCPUs, 128 GB memory, 15 Gbps network, 4000 GiB EBS).",[48,1219,1220],{},"This configuration was necessary to work around MSK's storage bandwidth limitations, ensuring a comparable cost basis to other evaluated streaming engines.",[48,1222,1223],{},"Additional key assumptions include:",[449,1225,1226,1229,1232],{},[452,1227,1228],{},"Inter-AZ producer traffic: For leader-based engines, two-thirds of producer-to-broker traffic crosses AZs due to leader distribution.",[452,1230,1231],{},"Consumer optimizations: Follower fetch is enabled across all tests, eliminating inter-AZ consumer traffic.",[452,1233,1234],{},"Storage cost exclusions: This benchmark only evaluates streaming costs, assuming no long-term data retention.",[32,1236,1238],{"id":1237},"inter-broker-replication-costs","Inter-Broker Replication Costs",[48,1240,1241],{},"Inter-broker (cross-AZ) replication is a major cost driver for data streaming engines:",[449,1243,1244,1247,1250],{},[452,1245,1246],{},"RedPanda: Inter-broker replication is not free, leading to substantial costs when data must be copied across multiple availability zones.",[452,1248,1249],{},"AWS MSK: Inter-broker replication is free, but MSK instance pricing is significantly higher (e.g., $3.264 per hour for kafka.m7g.8xlarge vs $1.306 per hour for an on-demand m7g.8xlarge). The storage price of MSK is $0.10 per GB-month which is significantly higher than st1, which costs $0.045 per GB-month. Even though replication is free, client-to-broker traffic still incurs inter-AZ charges.",[452,1251,1252],{},"Ursa: No inter-broker replication costs due to its leaderless architecture, eliminating inter-zone replication costs entirely.",[32,1254,1256],{"id":1255},"zone-affinity-reducing-inter-az-costs","Zone Affinity: Reducing Inter-AZ Costs",[48,1258,1259],{},"We evaluated zone affinity mechanisms to further reduce inter-AZ data transfer costs.",[48,1261,1262],{},"Consumers:",[449,1264,1265],{},[452,1266,1267],{},"Follower fetch is enabled across all tests, ensuring consumers fetch data from replicas in their local AZ—eliminating inter-zone consumer traffic except for metadata lookups",[48,1269,1270],{},"Producers:",[449,1272,1273,1282,1291],{},[452,1274,1275,1276,1281],{},"Kafka protocol lacks an easy way to enforce producer AZ affinity (though ",[55,1277,1280],{"href":1278,"rel":1279},"https:\u002F\u002Fcwiki.apache.org\u002Fconfluence\u002Fdisplay\u002FKAFKA\u002FKIP-1123:+Rack-aware+partitioning+for+Kafka+Producer",[264],"KIP-1123"," aims to address this). And it only works with the default partitioner (i.e., when no record partition or record key is specified).",[452,1283,1284,1285,1290],{},"Redpanda recently introduced ",[55,1286,1289],{"href":1287,"rel":1288},"https:\u002F\u002Fdocs.redpanda.com\u002Fredpanda-cloud\u002Fdevelop\u002Fproduce-data\u002Fleader-pinning\u002F",[264],"leader pinning",", but this only benefits setups where producers are confined to a single AZ—not applicable to our multi-AZ benchmark.",[452,1292,1293,1294,1299],{},"Ursa is the only system in this test with ",[55,1295,1298],{"href":1296,"rel":1297},"https:\u002F\u002Fdocs.streamnative.io\u002Fdocs\u002Fconfig-kafka-client#eliminate-cross-az-networking-traffic",[264],"built-in zone affinity for both producers and consumers",". It achieves this by embedding producer AZ information in client.id, allowing metadata lookups to route clients to local-AZ brokers, eliminating inter-AZ producer traffic.",[32,1301,1303],{"id":1302},"cost-comparison-results","Cost Comparison Results",[48,1305,973],{},[449,1307,1308,1310],{},[452,1309,978],{},[452,1311,981],{},[48,1313,1314],{},"Ursa’s leaderless architecture, zone affinity, and native cloud storage integration deliver unparalleled cost efficiency, making it the most cost-effective choice for high-throughput data streaming workloads.",[48,1316,1317],{},[615,1318],{"alt":18,"src":1319},"\u002Fimgs\u002Fblogs\u002F679c72208198ca36a352f228_AD_4nXeeZuM8T-xBlD4Vf3j67K618n08qh8wIDLLtiLJG0ssA1Wj1V26u7wIDTX9sqLrtw8mB2c299dwzarGen62CG0Vh7nWstn5qbPGFcBaKJYEepTsLr5fHWv1U8uqbg8Y0UOK6fJ7.png",[48,1321,1322],{},[615,1323],{"alt":18,"src":1324},"\u002Fimgs\u002Fblogs\u002F679c625978031f40229de484_AD_4nXdLkLLJ30KKr-_A_rN1j8akVwBYacAWIPzWHoOReJF421890kfByZoQQxkLczihVSmiw5Q9J51-V9I2SEKITbwsYnANDDTlAVL5nQ_jfaHNTe9VEWhSoa7DZooCnilDYL6l6msmJg.png",[48,1326,1327],{},"The detailed infrastructure cost calculations for each data streaming engine are listed below:",[32,1329,1331],{"id":1330},"streamnative-ursa","StreamNative - Ursa",[449,1333,1334,1337,1340,1343],{},[452,1335,1336],{},"Server EC2 costs: 9 * $1.536\u002Fhr = $14",[452,1338,1339],{},"Client EC2 costs: 9 * $1.536\u002Fhr =$14",[452,1341,1342],{},"S3 write requests costs: 1350 r\u002Fs * $0.005\u002F1000r * 3600s = $24",[452,1344,1345],{},"S3 read requests costs: 1350 r\u002Fs * $0.0004\u002F1000r * 3600s = $2",[32,1347,1349],{"id":1348},"aws-msk","AWS MSK",[449,1351,1352,1355,1358],{},[452,1353,1354],{},"Server EC2 costs: 15 * $3.264\u002Fhr = $49",[452,1356,1357],{},"Client side EC2 costs: 9 * $1.536\u002Fhr =$14",[452,1359,1360],{},"Interzone traffic - producer to broker: 5GB\u002Fs * ⅔ * $0.02\u002FG(in+out) * 3600 = $240",[32,1362,1364],{"id":1363},"redpanda","RedPanda",[449,1366,1367,1369,1371,1374,1377],{},[452,1368,1336],{},[452,1370,1339],{},[452,1372,1373],{},"Interzone traffic - producer to broker: 5GB\u002Fs * ⅔ * $0.02\u002FGB(in+out) * 3600 = $240",[452,1375,1376],{},"Interzone traffic - replication: 10GB\u002Fs * $0.02\u002FGB(in+out) * 3600 = $720",[452,1378,1379],{},"Interzone traffic - broker to consumer: $0 (fetch from local zone)",[48,1381,1382,1383,1388],{},"Please note that we were unable to test ",[55,1384,1387],{"href":1385,"rel":1386},"https:\u002F\u002Fwww.redpanda.com\u002Fblog\u002Fcloud-topics-streaming-data-object-storage",[264],"Redpanda with Cloud Topics",", as it remains an announced but unreleased feature and is not yet available for evaluation. Based on the limited information available, while Cloud Topics may help optimize inter-zone data replication costs, producers still need to traverse inter-availability zones to connect to the topic partition owners and incur inter-zone traffic costs of up to $240 per hour.",[449,1390,1391,1397],{},[452,1392,1393,1396],{},[55,1394,1280],{"href":1278,"rel":1395},[264]," (when implemented) will help mitigate producer-to-broker inter-zone traffic, but it is not yet available. And it only works with the default partitioner (no record partition or key is specified).",[452,1398,1399],{},"Redpanda’s leader pinning helps only when all producers for the pinned topic are confined to a single AZ. In multi-AZ environments (like our benchmark), inter-zone producer traffic remains unavoidable.",[48,1401,1402],{},"Additionally, Redpanda’s Cloud Topics architecture is not documented publicly. Their blog mentions \"leader placement rules to optimize produce latency and ingress cost,\" but it is unclear whether this represents a shift away from a leader-based architecture or if it uses techniques similar to Ursa’s zone-aware approach.",[48,1404,1405],{},"We may revisit this comparison as more details become available.",[40,1407,1409],{"id":1408},"comparing-total-cost-of-ownership","Comparing Total Cost of Ownership",[48,1411,1412],{},"As highlighted earlier, with a BYOC Ursa setup, you can achieve 5 GB\u002Fs throughput at just 5% of the infrastructure cost of a traditional leader-based data streaming engine, such as Kafka or RedPanda, while managing the infrastructure yourself. This significant cost reduction is enabled by Ursa’s leaderless architecture and lakehouse-native storage design, which eliminate overhead costs such as inter-zone traffic and leader-based data replication. By leveraging a lakehouse-native, leaderless architecture, Ursa reduces resource requirements, enabling you to handle high data throughput efficiently and at a fraction of the cost of RedPanda.",[48,1414,1415],{},"Now, let’s examine the total cost comparison, evaluating Ursa alongside other vendors, including those that have adopted a leaderless architecture (e.g., Confluent WarpStream). This comparison is based on a 5GB\u002Fs workload with a 7-day retention period, factoring in both storage cost and vendor costs Here are the key findings:",[449,1417,1418,1421,1424],{},[452,1419,1420],{},"Ursa ($164,353\u002Fmonth) is: 50% cheaper than Confluent WarpStream ($337,068\u002Fmonth)",[452,1422,1423],{},"85% cheaper than AWS MSK ($1,115,251\u002Fmonth)",[452,1425,1426],{},"86% cheaper than Redpanda ($1,202,853\u002Fmonth)",[48,1428,1429],{},"In addition to Ursa’s architectural advantages—eliminating most inter-AZ traffic and leveraging lakehouse storage for cost-effective data retention—it also adopts a more fair and cost-efficient pricing model: Elastic Throughput-based pricing. This approach aligns costs with actual usage, avoiding unnecessary overhead.",[48,1431,1432],{},"Unlike WarpStream, which charges for both storage and throughput, Ursa ensures that customers only pay for the throughput they actively use. Ursa’s pricing is based on compressed data sent by clients, meaning the more data compressed on the client side, the lower the cost. In contrast, WarpStream prices are based on uncompressed data, unfairly inflating expenses and failing to incentivize customers to optimize their client applications.",[48,1434,1435],{},"This distinction is crucial, as compressed data reduces both storage and network costs, making Ursa’s pricing model not only more cost-effective but also more transparent and predictable.",[48,1437,1438],{},[615,1439],{"alt":18,"src":1440},"\u002Fimgs\u002Fblogs\u002F679c602d194800c9206d9d58_AD_4nXcFlf755xgyz7htxhMhBV5fGrsxy642mQNodt61DTok_z1dwkw5A6lkO5hatXVneCaB0anbZPAyvLI3MlIMuQEYLEACHHvQMOr5UfaB37dfzkdqewDEvcT-20VGd_zzvJsuA00zGA.png",[48,1442,1443],{},[615,1444],{"alt":18,"src":1445},"\u002Fimgs\u002Fblogs\u002F679c62594e9c2e629fae73aa_AD_4nXeU6cOgItnjLsEZCOf13TEvMY_SHWWIxYP2OYUj-B1GUPyWO78OG08K_v03hwYSVcg06f9dqDiGmdwy76vynjmiDGL5bluZ5_XF4nSU_r59oOZdfViXndXt6s11vVOY7qwfZN8v.png",[32,1447,1449],{"id":1448},"cost-breakdown","Cost Breakdown",[387,1451,1452],{"id":1330},"StreamNative – Ursa",[449,1454,1455,1458,1461,1464,1467],{},[452,1456,1457],{},"EC2 (Server): 9 × $1.536\u002Fhr × 24 hr × 30 days = $9,953.28",[452,1459,1460],{},"S3 Write Requests: 1,350 r\u002Fs × $0.005\u002F1,000 r × 3,600 s × 24 hr × 30 days = $17,496",[452,1462,1463],{},"S3 Read Requests: 1,350 r\u002Fs × $0.0004\u002F1,000 r × 3,600 s × 24 hr × 30 days = $1,400",[452,1465,1466],{},"S3 Storage Costs: 5 GB\u002Fs × $0.021\u002FGB × 3,600 s × 24 hr × 7 days = $63,504",[452,1468,1469],{},"Vendor Cost: 200 ETU × $0.50\u002Fhr × 24 hr × 30 days = $72,000",[387,1471,1473],{"id":1472},"warpstream","WarpStream",[449,1475,1476,1479],{},[452,1477,1478],{},"Based on WarpStream’s pricing calculator (as of January 29, 2025), we assume a 4:1 client data compression ratio, meaning 20 GB\u002Fs of uncompressed data translates to 5 GB\u002Fs of compressed data.",[452,1480,1481,1482,1487],{},"It's important to note that WarpStream’s pricing structure has fluctuated frequently throughout January. We observed the cost reported by their calculator changing from $409,644 per month to $337,068 per month. This variability has been previously highlighted in the blog post “",[55,1483,1486],{"href":1484,"rel":1485},"https:\u002F\u002Fbigdata.2minutestreaming.com\u002Fp\u002Fthe-brutal-truth-about-apache-kafka-cost-calculators",[264],"The Brutal Truth About Kafka Cost Calculators","”. To ensure transparency, we have documented the pricing as of January 29, 2025.",[48,1489,1490],{},[615,1491],{"alt":18,"src":1492},"\u002Fimgs\u002Fblogs\u002F679c602e42713e0028e9af5e_AD_4nXcu5_VWTLu9jRYs6zX1MBAOtLQEo5gyfNSWPcbpnQHXTa8qNCFAXezRR2E8daygzYTTwd4dhJjaLaLM8C6y_3OGbu2NS7pdvEv3a8-ptNKOg7AeKnYqPQCAYvQ5EuxzuI3JYIvY.png",[387,1494,1496],{"id":1495},"msk","MSK",[449,1498,1499,1502,1505],{},[452,1500,1501],{},"EC2 (Server): 15 * $3.264\u002Fhr × 24 hr × 30 days = $35,251",[452,1503,1504],{},"Interzone Traffic (Client-Server): 5 GB\u002Fs × ⅔ × $0.02\u002FGB (in+out) × 3,600 s × 24 hr × 30 days = $172,800",[452,1506,1507],{},"Storage: 5 GB\u002Fs × $0.1\u002FGB-month × 3,600 s × 24 hr × 7 days * 3 replicas = $907,200",[387,1509,1364],{"id":1510},"redpanda-1",[449,1512,1513,1516,1518,1521,1524],{},[452,1514,1515],{},"EC2 (Server): 9 × $1.536\u002Fhr × 24 hr × 30 days = $9953",[452,1517,1504],{},[452,1519,1520],{},"Interzone Traffic (Replication): 5 GB\u002Fs × 2 × $0.02\u002FGB (in+out) × 3,600 s × 24 hr × 30 days = $518,400",[452,1522,1523],{},"Storage: 5 GB\u002Fs × $0.045\u002FGB-month(st1) × 3,600 s × 24 hr × 7 days * 3 replicas = $408,240",[452,1525,1526],{},"Vendor Cost: $93,333 per month (based on limited information. See additional notes below).",[387,1528,1530],{"id":1529},"additional-notes","Additional Notes",[449,1532,1533],{},[452,1534,1535,1536,1541],{},"Redpanda does not publicly disclose its BYOC pricing, making it difficult to accurately assess its total costs. We refer to information from the whitepaper “",[55,1537,1540],{"href":1538,"rel":1539},"https:\u002F\u002Fwww.redpanda.com\u002Fresources\u002Fredpanda-vs-confluent-performance-tco-benchmark-report#form",[264],"Redpanda vs. Confluent: A Performance and TCO Benchmark Report by McKnight Consulting Group.","” for estimation purposes. Based on the Tier-8 pricing model in the whitepaper,  the estimated cost to support a 5GB\u002Fs workload would be $1.12 million per year ($93,333 per month). However, since this calculation is based on an estimation, we will revisit and refine the cost assessment once Redpanda publishes its BYOC pricing.",[48,1543,1544],{},[615,1545],{"alt":18,"src":1546},"\u002Fimgs\u002Fblogs\u002F679c602dc8a9859eed89a0ef_AD_4nXdbcO8vsNNPy4GtkNLlmNKf22fjxRvzLzH7CtOna1L08sTbvnZx3HhufeFqc1w4K2gEF7lxO2IR5supotxebAiGnA07Qa8Yr3Rd1pVK2LYKK4WurlJGwgdwwucZIFoF-N_2oBjY.png",[48,1548,1549],{},[615,1550],{"alt":18,"src":1551},"\u002Fimgs\u002Fblogs\u002F679c602d6bc1c2287e012540_AD_4nXfcHZnLfjbjIr3ZAgoQXT9dwP3aQCOQPmGZZJUtpNZSwE6qY6M3yehIaBxCwxEIeu5PVdUPY0zhyjnow26YfgjdYgSG4GnV9ibxu0YWTIpwng6z_F6FUGJMpERMKtpsFESzXSN_Sw.png",[449,1553,1554,1557],{},[452,1555,1556],{},"When estimating the storage costs for Kafka and Redpanda, we assume the use of HDD storage at $0.045\u002FGB, based on the premise that both systems can fully utilize disk bandwidth without incurring the higher costs associated with GP2 or GP3 volumes. However, in practice, many users opt for GP2 or GP3, significantly increasing the total storage cost for Kafka and Redpanda.",[452,1558,1559],{},"Unlike disk-based solutions, S3 storage does not require capacity preallocation—Ursa only incurs costs for the actual data stored. This contrasts with Kafka and Redpanda, where preallocating storage can drive up expenses. As a result, the real-world storage costs for Kafka and Redpanda are often 50% higher than the estimates above.",[40,1561,849],{"id":848},[48,1563,1564],{},"Ursa represents a transformative shift in streaming data infrastructure, offering cost efficiency, scalability, and flexibility without compromising durability or reliability. By leveraging a leaderless architecture and eliminating inter-zone data replication, Ursa reduces total cost of ownership by over 90% compared to traditional leader-based streaming engines like Kafka and Redpanda. Its direct integration with cloud storage and scalable metadata & index management via Oxia ensure high availability and simplified infrastructure management.",[32,1566,1568],{"id":1567},"balancing-latency-and-cost","Balancing Latency and Cost",[48,1570,1571,1575],{},[55,1572,1574],{"href":1573},"\u002Fblog\u002Fcap-theorem-for-data-streaming","Ursa trades off slightly higher latency for ultra low cost",", making it an ideal choice for the majority of streaming workloads, especially those that prioritize throughput and cost savings over ultra-low latency. Meanwhile, StreamNative’s BookKeeper-based engine remains the preferred solution for real-time, latency-sensitive applications. By combining these two approaches, StreamNative empowers customers with the flexibility to choose the right engine for their specific needs—whether it's maximizing cost savings or achieving ultra low-latency real-time performance.",[32,1577,1579],{"id":1578},"the-future-of-streaming-infrastructure","The Future of Streaming Infrastructure",[48,1581,1582],{},"In an era where data fuels AI, analytics, and real-time decision-making, managing infrastructure costs is critical to sustaining innovation. Ursa is not just a cost-cutting alternative—it is a forward-thinking, lakehouse-native platform that redefines how modern data streaming infrastructure should be built and operated.",[48,1584,1585,1586,1591],{},"Whether your priority is reducing costs, improving flexibility, or ingesting massive data into lakehouses, Ursa delivers a future-proof solution for the evolving demands of real-time data streaming. ",[55,1587,1590],{"href":1588,"rel":1589},"https:\u002F\u002Fconsole.streamnative.cloud\u002F",[264],"Get started"," with StreamNative Ursa today!",[1593,1594,1596],"h1",{"id":1595},"references","References",[48,1598,1599,895,1601],{},[892,1600,1063],{},[55,1602,1603],{"href":1603},"\u002Fblog\u002Fintroducing-oxia-scalable-metadata-and-coordination",[48,1605,1606,895,1608],{},[892,1607,1012],{},[55,1609,1011],{"href":1011},[48,1611,1612,895,1615],{},[892,1613,1614],{},"StreamNative pricing",[55,1616,1617],{"href":1617,"rel":1618},"https:\u002F\u002Fdocs.streamnative.io\u002Fdocs\u002Fbilling-overview",[264],[48,1620,1621,895,1624],{},[892,1622,1623],{},"WarpStream pricing",[55,1625,1626],{"href":1626,"rel":1627},"https:\u002F\u002Fwww.warpstream.com\u002Fpricing#pricingfaqs",[264],[48,1629,1630,895,1633],{},[892,1631,1632],{},"AWS S3 pricing",[55,1634,1635],{"href":1635,"rel":1636},"https:\u002F\u002Faws.amazon.com\u002Fs3\u002Fpricing\u002F",[264],[48,1638,1639,895,1642],{},[892,1640,1641],{},"AWS EBS pricing",[55,1643,1644],{"href":1644,"rel":1645},"https:\u002F\u002Faws.amazon.com\u002Febs\u002Fpricing\u002F",[264],[48,1647,1648,895,1651],{},[892,1649,1650],{},"AWS MSK pricing",[55,1652,1653],{"href":1653,"rel":1654},"https:\u002F\u002Faws.amazon.com\u002Fmsk\u002Fpricing\u002F",[264],[48,1656,1657,895,1660],{},[892,1658,1659],{},"The Brutal Truth about Kafka Cost Calculators",[55,1661,1484],{"href":1484,"rel":1662},[264],[48,1664,1665,895,1668],{},[892,1666,1667],{},"Redpanda vs. Confluent: A Performance and TCO Benchmark Report by McKnight Consulting Group",[55,1669,1538],{"href":1538,"rel":1670},[264],{"title":18,"searchDepth":19,"depth":19,"links":1672},[1673,1674,1675,1680,1684,1685,1694,1697],{"id":352,"depth":19,"text":970},{"id":1005,"depth":19,"text":1006},{"id":1030,"depth":19,"text":1031,"children":1676},[1677,1678,1679],{"id":1042,"depth":279,"text":1043},{"id":1067,"depth":279,"text":1068},{"id":1088,"depth":279,"text":1089},{"id":1112,"depth":19,"text":1113,"children":1681},[1682,1683],{"id":1116,"depth":279,"text":1117},{"id":1131,"depth":279,"text":1132},{"id":1172,"depth":19,"text":1173},{"id":1184,"depth":19,"text":1185,"children":1686},[1687,1688,1689,1690,1691,1692,1693],{"id":1191,"depth":279,"text":1192},{"id":1237,"depth":279,"text":1238},{"id":1255,"depth":279,"text":1256},{"id":1302,"depth":279,"text":1303},{"id":1330,"depth":279,"text":1331},{"id":1348,"depth":279,"text":1349},{"id":1363,"depth":279,"text":1364},{"id":1408,"depth":19,"text":1409,"children":1695},[1696],{"id":1448,"depth":279,"text":1449},{"id":848,"depth":19,"text":849,"children":1698},[1699,1700],{"id":1567,"depth":279,"text":1568},{"id":1578,"depth":279,"text":1579},"2025-01-31","Discover how Ursa achieves 5GB\u002Fs Kafka workloads at just 5% of the cost of traditional streaming engines like Redpanda and AWS MSK. See our benchmark results comparing infrastructure costs, total cost of ownership (TCO), and performance across leading Kafka vendors.","\u002Fimgs\u002Fblogs\u002F679c6593d25099b1cdcec4ca_image-31.png",{},"\u002Fblog\u002Fhow-we-run-a-5-gb-s-kafka-workload-for-just-50-per-hour","30 min",{"title":945,"description":1702},"blog\u002Fhow-we-run-a-5-gb-s-kafka-workload-for-just-50-per-hour",[1710,1711,303],"TCO","Apache Kafka","A0o_2xdJiLI6rf6xj4RKsxJNo_A6QN2fYzCp6gaLrFw",[1714,1735],{"id":1715,"title":310,"bioSummary":1716,"email":290,"extension":8,"image":1717,"linkedinUrl":290,"meta":1718,"position":1732,"stem":1733,"twitterUrl":290,"__hash__":1734},"authors\u002Fauthors\u002Fxiangying-meng.md","Xiangying Meng is a Platform Engineer Intern at StreamNative.","\u002Fimgs\u002Fauthors\u002Fxiangying-meng.webp",{"body":1719},{"type":15,"value":1720,"toc":1730},[1721,1723,1727],[48,1722,1716],{},[48,1724,1725],{},[55,1726],{"href":344},[48,1728,1729],{},"‍",{"title":18,"searchDepth":19,"depth":19,"links":1731},[],"Platform Engineer Intern, StreamNative","authors\u002Fxiangying-meng","FAkNHd8ZxK1z8QFlBclJoDukSLP-8RCBSJPe5hqQ5y0",{"id":1736,"title":311,"bioSummary":1737,"email":290,"extension":8,"image":1738,"linkedinUrl":290,"meta":1739,"position":1746,"stem":1747,"twitterUrl":290,"__hash__":1748},"authors\u002Fauthors\u002Flishen-yao.md","Lishen Yao is a Software Engineer at StreamNative.","\u002Fimgs\u002Fauthors\u002Flishen-yao.png",{"body":1740},{"type":15,"value":1741,"toc":1744},[1742],[48,1743,1737],{},{"title":18,"searchDepth":19,"depth":19,"links":1745},[],"Software Engineer, StreamNative","authors\u002Flishen-yao","xd14od_8JPLjU2zdZkJ8vKS_Wb5EKKbFTuuA9CXGd9k",[1750,1758,1762],{"path":1751,"title":1752,"date":1753,"image":1754,"link":-1,"collection":1755,"resourceType":1756,"score":1757,"id":1751},"\u002Fsuccess-stories\u002Fbestpay","How Orange Financial combats financial fraud in over 50M transactions a day using Apache Pulsar","2022-12-27","\u002Fimgs\u002Fsuccess-stories\u002F67956b0055b8586d148f8b68_SN-SuccessStories-bestpay.webp","successStories","Case Study",1.1,{"path":1759,"title":1760,"date":1753,"image":1761,"link":-1,"collection":1755,"resourceType":1756,"score":1757,"id":1759},"\u002Fsuccess-stories\u002Ftencent","Powering Tencent Billing Platform with Apache Pulsar","\u002Fimgs\u002Fsuccess-stories\u002F67942e3b5f9411fb93dcb9f9_SN-SuccessStories-tencent.webp",{"path":1763,"title":1764,"date":1765,"image":1766,"link":-1,"collection":1767,"resourceType":1768,"score":1757,"id":1763},"\u002Fwhitepapers\u002Fapache-pulsar-helps-tencent-process-financial-transactions","Apache Pulsar Helps Tencent Process Tens of Billions of Financial Transactions Efficiently with Virtually No Data Loss","2022-12-23","\u002Fimgs\u002Fwhitepapers\u002F63aecf471d9c96089c87bff2_open-graph-wp-pulsar-tencent-billing.jpg","whitepapers","Whitepaper",1775716409415]