EMC doc "Using EMC Celerra with VMware vSphere and VMware Infrastructure" suggests some best practies to increase resiliency of virtual machines in case of failover of data movers. Other than increasing some timeouts, it says to change default SCSI driver of Windows 2003 systems with LSI Storport SCSI driver. The LSI Storport SCSI driver has architectural enhancements that provide performance improvements and avoid SCSI timeout errors/symmpi errors in Windows event viewer.
Is it a common practice? Is it still valid for Windows2008?
We didn't experience any of the SCSI timeout errors that the article refers to with the native drivers when failing over datamovers but that could have been more good luck than good management Definitely agree with the timeout and would have changed the drivers as well but decided to go with the "its not broke so why fix" approach.
Test and if you get the errors then upgrade.
In regards to Windows 2008 the timeout is still valid but changing the driver is not (refer to Table 2, page 320 of the EMC doc)