Ask the Expert: VMAX FAST VP and Performance

So what would be the impact if 1% FC allocation get filled up quickly ? Do you think new writes should be allocated from SATA even that 1% FC get full ? New writes will be sent to SATA

EMC has set subscription rate to zero on SATA Pool, so we cannot bind any devices to SATA Pool

Ok..

So FC it is has to be . The 1% Fc should ok on the basis that bind by policy is enabled.

Kevin - thanks for your response..Probably this would be my last question..

What if I remove even that 1% FC and I create FAST Policy with only SATA in it..Do you see anything bad in it except data cannot move to FC Tier ?

You cannot have a FAST policy with one tier.

If the SG is bound to FC it will likely block you from removing the 1% FC.

If you do not want FAST to work on the SG then removed it from FAST control.

It isn't so much putting it under FAST control as tagging it with a FAST VP policy as an effective way of identifying the intended performance you expect to see out of it. In our environment almost everything is under FAST VP control. If we leave an SG unassociated there is no way to tell if that is a mistake or if that was intended. Tagging it with a policy identifies it as an intentional act with a specific purpose.

You can have policy with only one tier, but what is the point. There will be no movements so FAST is essentially out of the equation. If an SG is bound to SATA and workload only has a SATA  requirement, and the application owners are happy with a SATA response time. Then why the extra management by putting in under FAST control.

I actually want to create a new fast policy with only SATA Tier in it because my intention is to free FC space by moving the data to SATA Pool..

Can't we create a new FAST policy with only 1 Tier in it ?

So if the customer determines that the workload only requires SATA and there is an issue with performance on that one application load then isn't that something the customer has accepted?

In addition I have to disagree with you on your statement that "You cannot create a FAST VP policy with one tier.' We have been doing ti for over a year with the EFD tier. Yes, I know we could do the same thing by binding directly to EFD (which we do) and not putting the SG under FAST control. But this way we apply a FAST VP policy that is only 100% EFD and we can clearly see on the list of SGs that that particular SG is pure EFD. Without the policy there is no easy way to see that the SG is pure EFD.

You won't be able to move to a pure SATA policy while the devs are bound to FC. There is a rule in the system that requires at least 1% allocation in a policy to the initially bound tier. We ran into this at one point and there isn't an easy way around it. Someone from EMC might be able to confirm but I'm pretty sure we tried rebinding and still found that the initially bound tier was required in the policy.

But as I said before my LUNS are bound to FC pool and I want the data to move to SATA pool but VLUN migration is not an option for us..so I am using SATA only FAST policy to bring the data from FC to SATA pool

Allen, thank you so much for highlighting this point..this is what exactly

I am looking for and I just found the similar condition in FAST Theory and

BP guide

Below are the lines from that doc :

When associating a storage group to a policy containing VP tiers, the

bound thin devices in the group must be bound to at least one of the thin

pools contained within the policy’s tiers.

A storage group associated with a FAST VP policy may contain unbound

devices. However, those devices may then only be bound to a thin pool

contained within the tiers in the policy

Thanks and Regards,

What is the command structure to enable bind by policy on FAST VP?

I would have to look up the SE command to do it (and I don't have those docs handy at the moment). You can do it through the Unishpere for VMAX GUI as well. Just go into the FAST dashboard and edit the FAST settings in the top left pane. You will have to show the advanced settings and then you will see a checkbox for it.

symfast -sid SymmID [-i Interval] [-c Count]

set -control_parms

[-vp_allocation_by_fp ]

-vp_allocation_by_fp — When set to ENABLE, for all thin devices managed by

FAST, Enginuity will choose a pool from the policy when making an allocation for the

thin device. When set to DISABLE, the allocation will be from the bound pool. The

default value for this option is DISABLE.

Note: The -vp_allocation_by_fp option is only supported by Enginuity 5876 and

1. higher. All Enginuity versions prior to 5876 will display N/A for this option.