Good Morning Everyone,
I'm assisting someone with a question on the Unisphere for VMAX Tiers Demand Report, which is accessible in the Unisphere UI on a Symmetrix array by navigating to [Array SID] > Storage > FAST. The Tiers Demand Report is shown as part of this dashboard, and specifically, we're looking at the report after maximizing it and switching from a Chart view to a Table view (via the drop-down menu at the top of the report window - please see attached for an example screenshot).
In this case, we're trying to determine how the "Available (GB)" metric is calculated, since it doesn't seem to correlate with the "Used (GB)" or "Free (GB)" metrics on the same report.
The documentation (Unisphere for VMAX Online help, p. 173-174) states the following, however it just isn't specific enough to answer our question:
Tiers Demand Report
Allows you view the demand on each tier in the Symmetrix system, as either a chart or a table.
The chart format provides the following information:
Used — The amount of storage that has already been used on the tier, in GB.
Free — The amount of free/unused storage on the tier, in GB.
Max SG Demand — The maximum amount of storage that FAST could put on the tier, in GB.
Available — The amount of storage available to FAST on the tier, in GB.
The table format provides the same information as the chart format, with the following additions:
Tech + Port — The type of disk on which the tier resides and the tier's RAID protection level.
Excess — The amount of excess storage available if FAST reaches the maximum of all the percentages in the policies associated with storage groups on the tier, in GB. This value can be either positive or negative. A positive value indicates that there will be enough excess storage available for all the storage groups. A negative value indicates that there will not be enough excess storage available and storage groups will be competing against one another.
All that being said:
*Could the team please help shed some light on how the "Available (GB)" column is calculated?
*Also, is there a Solutions Enabler command (or set of commands) we could use to also collect or prove out this information?
Many thanks in advance for your help & Kind Regards,
The "FAST Available" field shows the amount of capacity in a particular Tier that FAST VP is able to use. It's basically the total enabled capacity for all pools in a particular tier (the sum of the capacity of all enabled TDATs), minus the pool reserved capacity (PRC).
You can see the same report with SYMCLI with 'symfast -sid xxx list -demand -tech all -vp'.
Thank you very much for the quick response Sean!
This is excellent information and I really appreciate it. I'll pass those details along and see if it answers their questions OK(what you're saying makes sense to me, so hopefully that will clear things up), and will keep you posted.
Kind Regards & Thanks,
Hi Sean (and everyone),
Thanks again for the help, and if I could trouble you for just one or two more comments on this, I'd be in your debt.
Working off of the information you provided (thanks again for that), the person I'm assisting is still observing that in the Tiers Demand report, Available (GB) for a tier (vtier0, in this case) is actually larger than the pool (vtier0) itself. To my knowledge as well, there is only one pool which makes up this tier, hence the question.
In our example (please see the attached), vtier0's Free GB + Used GB = 5502GB, yet Available (GB) is 6406GB, and they are wondering why that is, since they've also stated that Pool Reserved Capacity (PRC) = 0GB.
I'm thinking that perhaps the discrepancy exists because of the use of Thin Devices (TDEVs) and oversubscription in the pool? However I'm not quite sure and I'm afraid my research has turned up empty - is there perhaps an easy way to view the amount of enabled capacity for the tier (which you indicated as the starting point for determining "Available (GB)" in the Tiers Demand report)?
Apologies if I've missed something small with regard to FAST or the report, but if you could lend me your thoughts, I'd really appreciate it!
Kind Regards & Thanks Again,
Something does seem off with the vtier0 report... there was a bug in the demand report in an earlier version of Solutions Enabler (7.6.0 I believe) -- I'm not sure if it affected Unisphere as well. My guess would be that you're either encountering this bug, or there's some stale data that needs to be refreshed in the SYMAPIDB or Unisphere. Might be a good idea to open an SR for this. I'd also check the SYMCLI demand report to see if it reports similar results... the SYMCLI command is 'symfast -sid <serial> list -demand -tech all -vp -v'.
Many thanks again for the lightning-fast response Sean!
I'll pursue that and see what I can find out (it's entirely possible that they may be running SE 7.6.0), and I'll also see if we can correllate the same via SymCLI. If we can, then I'll work on getting a Service Request opened with the SSG team.
OK, last question: Is there any chance you might know of a specific hotfix or version of SE where the bug you mentioned was fixed, so I could look it up?
Again thanks so much for the help & Kind Regards,
No problem Chris.
The fix is in SE 7.6.1 (and later)… here's the applicable text from the 7.6.1 release notes (in the "Fixed Problems" section):
"The calculation for FAST SG usage and FAST Avail is incorrect in the technology demand report."
Awesome - again this is exceptionally helpful information!
Thanks again for your help Sean (and to snalluri as well - I will look into the PRC value as well to make sure it's truly correct). I'll see what I can find out from here, and if needed, we'll open up a Service Request.
Much appreciated once again,
I was running thru' the same issue and good the issue has already been discussed. I will upgrade my SE soon.
In the FAST VP settings, "pool reserved capacity is set to default 10%", and If I would like to reduce the pool reserved capacity to minimum for the SSD pool where I'm not going to bound any tdev, shall I do that in the pool properties ?
So it will keep the PRC of FC pool to default 10% and SSD to 1%. Will it work?
In Unisphere for VMAX, yes -- it's in the pool properties screen.
For SYMCLI, you'd use symconfigure to configure the pool_resv_cap setting. For example --
symconfigure -sid 1234 -cmd "set pool Thin_Pool1, type=thin, pool_resv_cap=1;" commit
Ninja edit: Since you're not binding anything to the EFD pool, your proposed settings (10% global PRC, and 1% EFD pool PRC) are also correct.