Filer Volume size and snapshot

[hmakhlooq]

I have 2 Filers F840 with clustering anbled and both snapmiror and snapshot, thet are active s/by.

My problem is that I have a treditional volume with 1.78 TB size and 20% snapshot reservation. The volume shows 88% size utilization and snapshot 65%.

I deleted the snapshots ,but the volume size did not change (decrease). Also new snapshot of 10% created, but the volume size still smae.

The volume increase by 2% every two days. I would like to know why the size did not change after deleting the snapshot.

I deleted snapshot sometime back and it have effect on the total volume size, but now it did not.

Could anyone help??

some command run

filer-1> df
Filesystem              kbytes       used      avail capacity  Mounted on
/vol/myvolume/       2005250460 1752759636  252490824      87%  /vol/myvolume/
/vol/myvolume/.snapshot  501312612   46387116  454925496       9%

filer-1> df -r myvolume
Filesystem              kbytes       used      avail   reserved  Mounted on
/vol/myvolume/       2005250460 1752797864  252452596          0  /vol/myvolume/
/vol/myvolume/.snapshot  501312612   46402768  454909844          0 
filer-1>

Is it poosible to reduce the snapshot from 20% to 15 % without effecting the volume data??

[jb2cool01]

When you deleted this snapshot you should have noticed that the percentage used in the snap reserve went down. The available space in the rest of the volume would remain untouched though as snapshots live in the snap-reserve rather than the main volume.

To get more space available to you in the volume, you can shrink your snap-reserve from 20% to 15% or 10% or whatever is suitable for your environment. This will give you more space for data but less space available for snapshots.

[hmakhlooq]

Thanks Jason, do you mean 88% is the data space not include the snapshot??

To shrink the snapshot reserved space from 20% to 15%, wont effect the data on the tridditional volume. right?

Hussain

[jb2cool01]

Correct.

The volume is split into two parts. The main 'production' part where all your files sit and the snapshot reserve. That's why the volume space looks different when you look at it through Filerview and when you do a vol size command.

For instance, if you had a 1000Gb volume with a 50% snap reserve you'd only see 500Gb available to use (As the other 50% is reserved for snapshot data). If you shrink the snap reserve to 20% the available space in the production area would increase to 800Gb but you'd then only have 200Gb (20%) reserved for snapshot data. Depending on your snapshot schedule and your volume usage you might need a lot reserved for snapshots or you might not need much at all.

Make sure you read Anthony Feigl's reply though, that provides usefull info.

[hmakhlooq]

I agree with you, but my concern is that not effect the data after shrinking the snapshot. I can see Anthony Feigl saying it could effect the performance.

your comments please.

Hussain Makhlooq

[jb2cool01]

Whenever i have done it i have never noticed the difference (Probably because i did it so long ago).

My production snap reserves are set like this;

Root volume 10%

Home drives 10%

Scientific rata repositry 10%

General file storage 10%

Backup folder (Dumps from other systems) 50%

Watch out though. If your snap reserve was set at 20% and it was 65% full then shrinking the snap reserve will make the snap reserve more full. To me this volume looks like it just full and you need to expand it. Deleting snapshots and decreasing the snap reserve will work for now but the snapshots will build up again. You insist that you want to decrease the snap reserve you might want to look at your snapshot schedule you ensure you create fewer snapshots (As there will be less space available for them).

[anthonyfeigl]

Here is a snippet from the NetApp best practices.

TR 3437

Snapshot reserves. Snapshot copies are backups of how a volume looks at a particular point in time. The unique NetApp approach allows Snapshot copies to work almost instantaneously with very little impact on storage capacity. Later, a storage administrator or even an end user can recover data from the desired Snapshot copy. By default, Snapshot reserve space is set to 20% of the total volume capacity and also by default keeps a minimum number of weekly, daily, and hourly Snapshot copies online and available. To meet business needs, a customer might increase the number of Snapshot copies kept online, the frequency of Snapshot copies, or both.

• Snapshot copies provide a natively integrated and easy-to-use data protection utility that helps reduce impact on storage administrators by enabling end users to recover their own data.
• Protects against inadvertent file modification or deletion by making point-in-time backups available.
• NetApp recommends leaving Snapshot reserve and frequency at the default values or increasing the frequency and scheduling to meet business requirements.
• Rapidly changing data increases the sizes of Snapshot copies, resulting in more impact on reserve space.
• More frequent Snapshot copies also have more impact on reserve space.
• For either of these conditions, be sure to keep ample spare capacity in the Snapshot reserve.

Anthony

[anthonyfeigl]

Hey Hussain,

Unless the snapshot percentage itself is 100% or greater, you will not reclaim any "production" volume space after deleting the snap shots.

The only way to add to "production" space, as per the previous comment, is to change your snap reserve from 20%, however, that causes performance impacts in the striping ability of RAID disk sets on NetApp.

I have spent many years pushing the limits of production space on filers (50+ Trad Vols at a time), due to budget constraints in financial organizations and I have found it is not worth the hassle.  Especially in a 7x24 financial production environment.  NetApp filers have impressed me over eight years at how much you can hammer them, and squeeze out of them with no capacity. (98% Trad vols, with 2% snapshot and 3 nightlys)

Go to your management, and get a purchase order for more capacity, then build a new RAID (Flex/Trad) set, or add to the existing one and do a reallocate during downtime.

Hope this helps.

Anthony Feigl

[hmakhlooq]

Deat Anthony Feigl

I can set the snapshot to only one nightly, but my question is that do you recommend me to change snapshot reserved from 20% to 15% which can hold the one nightly snapshot. I am affraid from your previous comment about the stripping and the preformance. We are in our final project of replacing the existing storage with new one. I just want to make sure about the impact it may happen when I reduce snap reserved from 20% to 15%.

Do I have to disable all snapshot then reduce it? or does not need.

Note: This is triditional volume.

Hussain Makhlooq

[anthonyfeigl]

Hey Hussain,

I am not saying you cannot do it, and I am not saying that you will destroy any data by doing it.

It is more of a best practice to "not" start doing it as a workaround.

Workarounds quickly become an ongoing operational activity.

The performance change will probably be negligable, you can do it and I have done it for years in large environments.

Unless your filer is way over utilized, it will not be noticeable.

I just try to stick with best practices on NetApp gear so as not to introduce issues into production (financial) environments.

You simply have to enter the command "snap reserve VOLNAME NEWPERCENTSIZE"

>>Do I have to disable all snapshot then reduce it? or does not need.

No.

as my example below:

FILER> snap reserve asis_test2
Volume asis_test2: current snapshot reserve is 20% or 6081740 k-bytes.
FILER> snap reserve asis_test2 15
FILER> snap reserve asis_test2
Volume asis_test2: current snapshot reserve is 15% or 4561304 k-bytes.

I changed snap reserve from 20% to 15%

Hope this helps.

Anthony

[hmakhlooq]

Dear Anthony / Jason,

I would like to thank you for your helpful information, I have reduced the snap reserved to 15% and it was successful. I gained 5% space in the data volume.

I can managed my snapshot with the 15% so far.

Thanks again

Hussain Makhlooq