This document compares the storage performance of the EMC VNX 5100 and IBM DS5300 storage systems. Key findings include:
- The IBM DS5300 has more memory per controller (8GB vs 4GB for EMC VNX 5100) which impacts caching.
- Testing of RAID 10, 5 and 6 configurations show generally superior random I/O performance for EMC VNX 5100, while IBM DS5300 performs better on sequential reads/writes.
- Adding SSD caching to EMC VNX 5100 via FAST Cache significantly boosts performance, with SSDs achieving over 20,000 IOPS.
The document compares the storage performance of IBM DS4700, IBM DS5300, and IBM Storwize V7000 storage systems using various RAID configurations. Key findings include:
1) The IBM Storwize V7000 showed the best overall performance on all RAID types, with IOPS up to 15% higher than the IBM DS5300 and significantly higher than the older IBM DS4700.
2) The IBM Storwize V7000 had significantly lower response times than the other systems across most tests.
3) For a typical database workload with 80% reads, the performance difference between RAID 5 and RAID 10 was around 20-25% for the IBM Stor
The document summarizes performance tests conducted on an IBM FlashSystem 840 storage array.
Key findings from the tests include:
1) The array delivered over 500,000 IOPS in a test with a 100% random workload and 80% reads/20% writes using 4KB blocks.
2) Maximum performance was achieved with a 60-70% read/30-40% write workload mix rather than 100% reads.
3) Updating the array's firmware while under a 100K IOPS workload caused only a brief drop in performance and a slight increase in response times, demonstrating that firmware updates can be performed with minimal impact even under moderate load.
The document summarizes performance tests of an IBM FlashSystem 810 storage array connected directly to a host and virtualized through an IBM SVC. When connected directly, the FlashSystem 810 achieved maximum performance of 225k IOPS with 64KB and 128KB block sizes. However, when presented through the SVC, performance was limited to 154k IOPS due to high CPU load on the SVC nodes. Disabling caching on the SVC slightly improved IOPS but increased latency. The tests demonstrate that for this configuration, the SVC CPU is the bottleneck, rather than the FlashSystem 810 itself.
IBM Storwize V7000 Ultimate Performance EngOleg Korol
油
The IBM Storwize V7000 was able to achieve 53,000 IOPS with an average response time of 11ms in synthetic SPC-1 testing using 240 HDDs. This document examines whether similar real-world performance can be achieved on a Storwize V7000 supporting a production database workload consisting of OLTP and OLAP applications across 208 HDDs, without using SSDs. Charts of disk usage and response times over a day show the Storwize V7000 supporting over 50,000 IOPS and average response times under 5ms.
The document summarizes the results of performance tests of different network connection technologies in IBM Power Systems logical partitions (LPARs). Test 1 showed an IVE adapter connected LPAR achieved 0.83Gbps for a single thread and 0.92Gbps for 16 threads. Test 2 using virtual ethernet between Hypervisor-connected LPARs achieved 1.33Gbps and 1.65Gbps respectively. Test 3 using an IVE adapter directly between LPARs achieved the highest speeds of 1.75Gbps and 5.88Gbps. The conclusion is that for high performance without 10GbE infrastructure, an IVE adapter provides the best performance with lower CPU usage than virtual ethernet.