Dell EqualLogic FS7500 Technical Manual page 10

ports. While this is usually not a recommended best practice for production networks (due to
scalability limits), it provided the connectivity we needed for this test environment.
On our SAN network, we configured jumbo frames (MTU 9000). This is established best practice for
EqualLogic PS Series iSCSI SANs and is also required for the FS7500 SAN and interconnect network
ports (which share the same switches with the SAN). For our client network, we ran some initial tests
with both standard and jumbo frames but did not find a significant difference in our particular test
cases. We decided to use the standard MTU size of 1500, which is also the most common size for
typical client networks.
The EqualLogic PS Series iSCSI SAN consisted of six PS6000XV arrays. Each array contained 600GB
15K SAS drives and was configured for RAID 10. Initially we created two pools – one for arrays
dedicated to block storage and another dedicated for file storage (i.e. for the FS7500). Later, we
migrated all of our volumes into a single pool that contained three arrays. We ran our tests and then
added another array to the pool and gave the volumes time to rebalance across the new array in the
pool. We repeated this process until we had all six arrays in a single pool.
For SAN testing, we created sixteen 10GB volumes on the SAN and presented these to each of the 16
VMs as Raw Device Mode (RDM) volumes. For file I/O (CIFS and NFS), we created a file system mount
point on the FS7500 which was then mounted from each of the 16 clients. Our NAS reserve was 5TB
total, although by comparison, we only used a small portion of that capacity in our tests.
A Vdbench script was created that would generate I/O at the desired block size (or I/O request size)
and run for a specific amount of time. After the test was run, we collected the results and analyzed
them.
BP1016 6
Integrating the Dell EqualLogic FS7500 into an Existing SAN