VMAX FAST-VP with SRDF

Hello All,

I need to know if VMAX FAST policies can be applied to Thin SRDF devices (devices with RDF1+TDEV attribute). If yes, Are there any watchouts for applying FAST policies to SRDF thin devices and if anyone has implemented this, please let me know the performance of SRDF with and without FAST policies.

Thank you.

Tags (1)
0 Kudos
1 Reply
Highlighted
Sniper712
1 Copper

Re: VMAX FAST-VP with SRDF

In Enginuity 5875, FAST VP operated independently on each side of the SRDF link. Promotion and demotion decisions made by FAST VP were based on the actual workload seen by each individual device.

While an R1 device would typically undergo a read and write workload mix, the corresponding R2 device would only see a write workload. (In 5875, reads against the R1 are not reflected across the link.). A consequence of this was that the R2 device data may not be located on the same tier as the related data on the R1 device.

Enginuity 5876 introduces SRDF awareness for FAST VP. The performance metrics collected for R1 devices are periodically transmitted across the link to the corresponding R2 devices.

On the R2 device, the R1 performance metrics are merged with the actual R2 metrics. FAST VP takes into account the workload on the R1 device and then makes promotion and demotion decisions for the R2 device data.

SRDF coordination is enabled, or disabled, per storage group. The default is disabled.

http://www.emc.com/collateral/hardware/white-papers/h10497-enginuity5876-new-features-vmax-wp.pdf