Volume Size: Show Actual Data Foot Print

[richarde]

We have over 50,000 users using multiple 8.1.1 7-mode 6080HA controllers with almost 2PB of storage. We have multiple "Customers" residing on our systems that use CIFS, NFS, iSCSI and FC (and soon FCoE) in a SMT type environment using multistore.

What I need to know is if I can have the volumes report actual data footprint vs a deduplicated/compressed size?

What this will give me is the ability to tell our customers they must request more space because the actual data footprint consumed is over their allocated space. This will also allow me to set the volume size at 10TB (or whatever number) and if the data is 10TB and it deduplicates to 6TB that space can be returned to the Aggr for redistribution to another customer. Currently the volume gets this space back and the customer thinks they have 4TB still available to them in which they use.

My customers are very political and if they see a volume is 10TB and has 4TB available they will not accept that they are using 10TB unless the volume reports to them that they are using 10TB and are 100% full. Without this feature I will not be able to get the best Efficiencies out of my storage due to the Political nature of this customer.

I know Ops Manager and other tools have the ability to report actual data used but the customers are able to see down to their own volumes and some of them own their own vFilers which give further insight..

If I completely missed a vol option somewhere in the manuals I apologize in advance. I have been researching the internal Wikid's and have not seen anything that suggests this is a current feature. I think this would be a useful feature for most so called "Cloud" providers and would be extremely helpful for our "Customer Cloud".

I know that quotas will do this for my CIFS share but will not help me when my customers are using vFilers/Multistore or other protocols.

Thank you for your help and any feedback is welcome.

[richarde]

So another question. Has anyone every thought this would be useful  or think it might be useful now that you have thought about it?

[radek_kubka]

I am not sure whether I properly understand your question, but wouldn't df -s output provide all the info you are after?

Regards,
Radek

[richarde]

I want my volumes to never display the deduplicated/compressed size. I only want CIFS, NFS and vFiler to be able to see actual data footprint. Any space recovered via compression and/or deduplication would then be returned to the aggr not to the volume. This should be set on a per volume level via vol options.

Good?

[radek_kubka]

I see, a very unusual one methinks

One of the ways would be to script volume auto-shrink based on df -s output. That may confuse users even more - they can see how much free space they are entitled for, but at the same time they will scratch their heads why their volumes are shrinking.

I doubt there is any neat solution for shares - it's completely opposite with LUNs, where from a host side no savings are ever seen.

[richarde]

So I was told that with cifs you can put a quota on the entire volume and the share would show the actual data foot print vs the total size. So 10TB of data would show 10TB of used space. This works for my shares but I have some customers that have there own vFilers and it does not work on them. They can see everything and if they can see it, they use it.

If I could set the volumes assigned to these vfilers/vservers (I am plaining on going clustered ontap) to only show actual data footprint then I can restrict them but give them what they requested in the first place.

Yes you are correct, having a volume auto shrink would cause major political issues in this shop.

Thanks for the help.