pale shard
Hi everyone !
I have a questions regarding best practices on ONTAP System manager.
There is a rule on best practice page : https://docs.netapp.com/us-en/flexpod/healthcare/medical-imaging_best_practices.html#storage-layout
" For optimal storage performance, NetApp recommends that you have at least 10% free space available in an aggregate. "
The free space refers to the "used and reserved space" or the "physical used" space ?
Thanks in advance for your replies 🙂
nova rune
it's what's physically free on the aggr overall.
so if you have 100TiB aggr. leave ~10TiB free.
pale shard
Thanks for your answer,
Just to be perfectly clear, if I have a 100Tib aggregate,
95Tib used and reserved
50Tib physical used
45Tib reserved
5Tib avaliable
I should free 5 more Tib to be good ?
( I work with ONTAP System Manager release 9.12.1P2 )
nova rune
What platform are you running?
but there's such a thing as aggr wafl reserve -. typically 10% which changed to 5% in 9.12.1 for AFF systems.
pale shard
I've got a FAS8300 and a FAS8200.
nova rune
here's how it's laid out visually - https://kb.netapp.com/onprem/ontap/os/Space_Usage
there's a reserver, but you don't want to run super full.
pale shard
Thanks for the link.
My question is purely theoretical. I just want to be sure that the free space refers to the available space (= total physical space - used and reserved space).
The other possibility is that the free space refers to (total physical space - physical used space)
nova rune
"Physical Used Percentage" should be 10% free.
so Physical used space in the 100TiB example would be ~10TiB.
pale shard
Here is a screenshot of my aggregate.
Maybe my question is more clear with this information.
Which value has to be < 90% ?
analog schooner
df -A -h is probably easier to parse...
nova rune
or just aggr show
Your % is getting up there . depending on your data growth, about 75-80% full is when i typically recomend planning for what to do about needing additional space.
fallow crane
however, to be exact - the "physical used" is the number you don't want to be over 90% of 250TiB for a HDD system of this size
so physical capacity of 225TiB is the number you don't want to exceed without plans for expansion or data reduction. "Reserved" doesn't matter for the purposes of WAFL capacity exhaustion
it's not a "hard" rule - WAFL = Write Anywhere File Layout. You don't want it to have to look too hard (long) for the "anywhere" part of "write"
But if it gets too full, it can take longer than you'd hope, to delete things. So it's best to keep on top of it and try to keep it under 90%
pale shard
Thanks guys !
The results of the commands "df -A -h" and "aggr show" displayed the percentage =~ 84% for this aggregate.
So this value include the reserved space + the physical used. I can easily reduce the reserved space on my volumes so the % will be < 90% on all my aggregates.
Do you guys agree with my reasoning ?
analog schooner
no... what df -A -h is showing you is the actual space used... blocks in use... you need more disks... and from the response times, I'd get a couple of flashcache modules as well
the easiest way to manage all of this as long as you're just doing NAS protocols is no snap reserve, no space guarantees and just watch your aggregate space... adjust volume sizes as needed
and learn to communicate about a budget, hehe...