Infrastructure modernization decisions have many moving parts and require many technological choices. While HCI has long been touted as a means to reduce your overall infrastructure costs, factors such as overall cost, manageability, and scalability feed into the final equation and your ultimate investment decision.
It's like purchasing a new car. The initial price of the base model might seem reasonable, but as you begin to include the necessary features and packages, costs can go up rapidly. While our focus certainly isn't car sales, when it comes to helping you optimize your Hyperconverged Infrastructure (HCI) investment, having confidence in the benchmark performance of your decision is a crucial starting point.
We recently tested and demonstrated that VMware vSAN 8 on 4th Gen Intel® Xeon® Scalable processors with 32 cores had higher HCIBench throughput and lower latency than AMD EPYC Genoa processors with 32 cores.
Higher Throughput on Intel® Xeon® Gold 6448Y over AMD Genoa 9354 were both tested with 32 cores, which is a common core count for VMware deployments.
- +32% at 32k
- +30% at 16K
- +24% at 8k
Lower Latency on Intel® Xeon® Gold 6448Y over AMD Genoa 9354
- 75% Reduction at 32k
- 23% Reduction at 16k
At 32K, modernizing to 4th Gen Intel Xeon Scalable processors can give you 32% more throughput PLUS a 75% reduction in latency over AMD Genoa. I’m not sure what that might translate to when deciding what features to add to a new car, but when it comes to modernizing your infrastructure, choosing Intel is easy.
We are passionate about the competitive advantages of Intel processors compared to AMD; if you’d like to learn more about how Intel compares favorably in VMware vSAN environments, email: patryk.wolsza@intel.com.
Configuration Details:
AMD Genoa 9354: 4 Node, 2x AMD® Geona® 9354 Processor, 4x AS-2025HS-TNR, Total Memory 768GB (24x 32G DDR5-4800MHz), HyperThreading: Enable, Turbo: Enabled, NPS=1, LLC=off, BIOS: 1.4(ucode:0x0a101116), Storage (data): 9x 3.84 TB Solidigm D7-P5510 Series PCIe NVMe , Network devices: 1x Intel Ethernet E810CQDA2 E810-CQDA2, fw 4.2, at 100 GbE RoCE, Network speed: 100 GbE, OS/Software: VMware 8.0U1c, 22088125, vSAN ESA – Optimal default policy (RAID-5), Test by Intel as of 08/22/2023, using HCI Bench 2.8.1, FIO3.3. Throughput test 8k profile (I/O size 8k, Read percentage 70%, Random percentage 100%, latency target mode<10ms, #VMs per cluster 16, vCPU 4, vRAM 8, # data disks per VM 4, size of disk 50GB). Latency test 8k profile (I/O size 8k, Read percentage 70%, Random percentage 100%, #VMs per cluster 16, vCPU 4, vRAM 8, # data disks per VM 8, size of disk 50GB)
Intel Xeon Gold 6448Y: 4-node cluster, 2x Intel® Xeon® Gold 6448Y per node, 4x Server M50FCP2SBSTD, Total Memory 512 GB (16x DDR5 32GB 4800MHz), HyperThreading: Enable, Turbo: Enabled, NUMA noSNC, BIOS: SE5C741.86B.01.01.0004.2303280404 (ucode:0x2b0001b0), Storage (data): 9x 3.84 TB Solidigm D7-P5510 Series PCIe NVMe , Network devices: 1x Intel Ethernet E810CQDA2 E810-CQDA2, fw 4.2, at 100 GbE RoCE, Network speed: 100 GbE, OS/Software: VMware 8.0U1c, 22088125, vSAN ESA – Optimal default policy (RAID-5), Test by Intel as of 08/22/2023, using HCI Bench 2.8, FIO3.3. Throughput test 8k profile (I/O size 8k, Read percentage 70%, Random percentage 100%, latency target mode<10ms, #VMs per cluster 16, vCPU 4, vRAM 8, # data disks per VM 4, size of disk 50GB). Latency test 8k profile (I/O size 8k, Read percentage 70%, Random percentage 100%, #VMs per cluster 16, vCPU 4, vRAM 8, # data disks per VM 8, size of disk 50GB)
Notices and Disclaimer:
- Performance varies by use, configuration, and other factors. Learn more on the Performance Index site.
- Performance results are based on testing as of dates shown in configurations and may not reflect all publicly available updates. See backup for configuration details. No product or component can be absolutely secure.
- Your costs and results may vary.
- Intel technologies may require enabled hardware, software, or service activation.
- © Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries. Other names and brands may be claimed as the property of others.
Patryk is a vExpert and Cloud Solutions Architect in Intel Datacenter & AI Group, with a focus on Software Defined Infrastructure. With more than a decade of expertise in different cloud platforms, Patryk has broad experience in cloud solutions, influencing data center designs, and understanding connections between ordinary Data Centers, virtualization, SDI, cloud, and edge. His recent focus is on optimization workloads leveraging differentiation features and building technology-enabling stacks.