So, does anyone have a definition of Written Capacity?

[JBARBALACE]

I find this to be a rather curious entry in the published Inventory User Guide for OCI 7.2.2:

"Written Capacity JK: Need Ram's input"

https://library.netapp.com/ecm/ecm_download_file/ECMP12506920

Did Ram ever provide input? I'd like to better understand the difference between Written and Consumed Capacity in the Backend Volumes view.

Thank you,

Julia

[merdos]

I understand since (from Ketsia Pha) that in OCI Written is from the Volume perspective and Consumed is the ammount of capacity consumed in the aggr or storage pool, so any overhead would be included in Consumed and Consumed is expected to be equal or higer than Written Capacity. 

[ostiguy]

Well, that isn't ideal. I laughed, but that was probably inappropriate. I opened a documentation bug.

We only have this data for a few storage platforms - I think some SVC, and Vmax generation 1 and maybe 2. EMC seems to have de-emphasized this, and I don't think it is available for the Vmax 3 family.

A couple platforms of array purport to know exactly how much data has been written to a volume. This theoretically would line up more closely with the hosts' perspective on how much capacity utilization is occurring.

If OCI can obtain this for block volumes, it would be populated here.

[JBARBALACE]

Thank you for the quick follow-up and the bug report!

The capacity in the Backend Volumes view is coming from several VMAX. There is a VPLEX on the front end, so we don't see consumed, written, etc in the Volume view for host.

The block capacity is only reported in Backend, and to a lesser extent in the Virtual Storage View. Written Capacity is indeed less than Consumed.

To add to this, is there a reason why the Raw Capacity is double the Capacity?  I wondered if it was being duplicated because of the VPLEX, the Redundancy, something else?

Thanks again,

Julia

[ostiguy · Accepted solution]

Raw capacity for a volume being 2x its capacity would be what I expect for a RAID 1 or RAID 10 configuration - is that the case for these volumes?

[JBARBALACE]

That's where my thought was going. The redundancy reports as 2-way mirror, which I would expect to be RAID1.  I'll confirm with the storage engineers.