Aerospike’s Post

Learn key technology differences between #Aerospike and #Redis Enterprise in our database comparison. 👇 https://lnkd.in/gGEx5Bkw #RedisReplaced

Need the right NoSQL for your high-performance applications? Compare key capabilities of Aerospike 7.0 and Redis Enterprise 7.2 in our deep dive: https://lnkd.in/eHawsPfg #Aerospike #RealTimeData #NoSQL #TechComparison #Redis

Intel, AMD, or ARM? These are the best servers for Apache Kafka! 👇 Recently, I came across an in-depth benchmarking study conducted by DoubleCloud to evaluate the performance of Apache Kafka clusters in terms of cost-efficiency. Kafka is still the gold standard for message queues, especially for self-hosting. I found the research super insightful as it reveals how infrastructure costs correlate with the ability to ingest millions of rows into Kafka brokers, with a unique focus on records per cent—not just data throughput. This is a metric that almost no one else looks at. I was also very surprised by the results! If you're interested in a detailed look at the methodology and results, check out the full research article here: https://bit.ly/4bEDHb1 My key findings were: ✅ Old AMD M5 instances are bad: low throughput, high latency and relatively expensive ✅ Solid M6 Intel and AMD instances: good throughput, low latency, good price ✅ New M7 AMDs are fire🔥: best throughput, lowest latency but also most expensive ✅ ARM is the secret winner: good latency, awesome throughput and very cheap! ✅ Nearly 50% use ARM: DoubleCloud says that nearly 50% of their customers already use ARM clusters! It's super cool that DoubleCloud makes and shares these benchmarks. Often, only performance benchmarks are done. PS: Besides AWS, two GCP server types were also examined in more detail. So definitely worth a read! #dataengineering #datascience #apachekafka #aws #gcp #doublecloud

A great post by Jeff Chou from Sync Computing showing cost optimization of Databricks using fixed instances vs. autoscaling. Specifically "a fixed cluster could outperform an autoscaled cluster in both runtime and costs for the 3 workloads that we looked at by 37%, 28%, and 65%." DM me for a demo of how Speedata.io delivers 50x better performance for Apache Spark vs. CPUs and GPUs. https://lnkd.in/gYWP9gsh

Next blog post in the series up!

In this next post on Elasticsearch vs. OpenSearch benchmarking, we take a closer look at the full benchmark done by Elastic. Unfortunately, we discover that it likely says nothing about the relative performance. Why? A mix of queries matching no documents, caching and latency (probably network) issues.

Intel, AMD, or ARM? These are the best servers for Apache Kafka! 👇 Recently, I came across an in-depth benchmarking study conducted by DoubleCloud to evaluate the performance of Apache Kafka clusters in terms of cost-efficiency. Kafka is still the gold standard for message queues, especially for self-hosting. I found the research super insightful as it reveals how infrastructure costs correlate with the ability to ingest millions of rows into Kafka brokers, with a unique focus on records per cent—not just data throughput. This is a metric that almost no one else looks at. I was also very surprised by the results! If you're interested in a detailed look at the methodology and results, check out the full research article here: https://bit.ly/4bEDHb1 My key findings were: ✅ Old AMD M5 instances are bad: low throughput, high latency and relatively expensive ✅ Solid M6 Intel and AMD instances: good throughput, low latency, good price ✅ New M7 AMDs are fire🔥: best throughput, lowest latency but also most expensive ✅ ARM is the secret winner: good latency, awesome throughput and very cheap! ✅ Nearly 50% use ARM: DoubleCloud says that nearly 50% of their customers already use ARM clusters! It's super cool that DoubleCloud makes and shares these benchmarks. Often, only performance benchmarks are done. PS: Besides AWS, two GCP server types were also examined in more detail. So definitely worth a read! #dataengineering #datascience #apachekafka #aws #gcp #doublecloud

In large-scale MinIO deployments, it is not unusual to see server counts in the hundreds. Each of those servers has 500GB or more of DRAM, often it is a full TB. What if you could pool all that DRAM into a shared cache that could be used as an in-memory object store? Well, you can. Introducing MinIO #EnterpriseObjectStore Cache. It is a distributed, shared DRAM cache and is ideal for workloads that demand massive IOPS and throughput performance like….you guessed it—#AI applications. Check out the details here: https://lnkd.in/gr_8yrgm

Amazing performance improvement with usage of integrated cache . Do read more about it in the blog below. 1️⃣ Improvement in latencies. 2️⃣P75 latency is down 75% and 3️⃣ P99.9 latency is down over 67% 4️⃣ Reduced infrastructure from 60k cpu cores to 3k https://lnkd.in/g6QA4YVm

Intel, AMD, or ARM? These are the best servers for Apache Kafka! 👇 Recently, I came across an in-depth benchmarking study conducted by DoubleCloud to evaluate the performance of Apache Kafka clusters in terms of cost-efficiency. Kafka is still the gold standard for message queues, especially for self-hosting. I found the research super insightful as it reveals how infrastructure costs correlate with the ability to ingest millions of rows into Kafka brokers, with a unique focus on records per cent—not just data throughput. This is a metric that almost no one else looks at. I was also very surprised by the results! If you're interested in a detailed look at the methodology and results, check out the full research article here: https://bit.ly/4bEDHb1 My key findings were: ✅ Old AMD M5 instances are bad: low throughput, high latency and relatively expensive ✅ Solid M6 Intel and AMD instances: good throughput, low latency, good price ✅ New M7 AMDs are fire🔥: best throughput, lowest latency but also most expensive ✅ ARM is the secret winner: good latency, awesome throughput and very cheap! ✅ Nearly 50% use ARM: DoubleCloud says that nearly 50% of their customers already use ARM clusters! It's super cool that DoubleCloud makes and shares these benchmarks. Often, only performance benchmarks are done. Besides AWS, two GCP server types were also examined in more detail. So definitely worth a read! By the way, here you can explore DoubleCloud's managed service for Apache Kafka with a free trial: https://bit.ly/3WB7soX #dataengineering #datascience #apachekafka #aws #gcp #doublecloud