1dent1ty cHa0s

I've had some time recently to setup a test rig at home and begin performing some baseline performance tests of our biggest performance problem, the initial load experience with the FIM MA. Here is some information on my test system:

• Windows Server 2008 R2 Datacenter running Hyper-V
• Intel Core i5-760 2.8GHz Quad Core
• 16GB RAM (4 4GB DDR3 1333)
• EVGA P55V (120-LF-E651-TR) – Intel P55 1156 motherboard with onboard RAID (non-caching)
• 4 Seagate Barracuda 7200 RPM 1.5TB SATA II (3Gb/s) in RAID 5 array hosting the OS and VM’s
• 1 Samsung Spinpoint 5400 RPM 2TB SATA II (3Gb/s) as dedicated backup (Volume Shadow Copy) volume

Now, for a point of reference, this is probably the worst disk configuration you can have when it comes to SQL Server and FIM performance metrics. As we'll see the low disk I/O, RAID level and lack of RAID cache will really cause the numbers to fall on my initial test. Shortly after performing my initial test, I upgraded my test rig with the following components:

• LSI MegaRAID SAS 9260-4i 512MB Caching RAID controller
• OCZ Vertex 2 60GB SATA II SSD

The disk configuration for the subsequent tests is as follows:

• 1 OCZ Vertex 2 60GB SATA II SSD (3Gb/s) hosting the Hyper-V Host OS (AHCI with TRIM)
• 4 Seagate Barracuda 7200 RPM 1.5TB SATA II (3Gb/s) in RAID 10 array hosting the VM’s
• 1 Samsung Spinpoint 5400 RPM 2TB SATA II (3Gb/s) as dedicated backup (Volume Shadow Copy) volume

Hosting the Hyper-V host on the SSD really screams and the addition of the caching array controller running in RAID 10 mode made a measurable difference to the performance of all of the guest VM's. In future tests I'll try moving the database VHD's to the SSD.

The subsequent posts will focus on performance improvements through disk upgrades as well as the ones introduced in the new 4.0.3573.2 hotfix rollup.