Celerra iSCSI Luns and File System Sizing

Posted: August 11, 2009 in Celerra, VMware

Ive deployed a number of Celerra’s into customer sites over the last year and although the feature packed Celerra can present storage using CIFS and NFS, the majority of systems have been deployed using iSCSI for VMware environments.

Now if your not planning to replicate or snapshot your iSCSI luns, then these considerations do not apply, the additional file system space needed is only a fraction of the lun size to store meta data.

If your familiar with the Celerra then you’ll know that each iSCSI lun sits inside a Celerra file system, While it is possible to have multiple iSCSI luns inside a single file system, its not a good idea especially if you plan to snapshot your iSCSI luns. So with that said, the actual sizing considerations are actually around the size of the Celerra file system which will be used to store A. The Lun  B. The Snapshots and C.The temporary writable snapshots.

The size of the file system needed to host each iSCSI lun really depends on how you choose to deploy your iSCSI luns, lets look at each of these options and begin by defining a general equation.

 File System minimum = Size of Lun + Snapshots + Size of TWS (Temporary Writable Snapshot)

Here is a table showing the variables we will use later on during the sizing examples.

var

   

 

Fully Provisioned iSCSI Lun with Fully Provisioned TWS

thick

Celerra allocates enough space for a 100 percent data change (A) during the initial snapshot as a precaution. When the next snapshot is created, additional space is allocated equivalent to the amount of changed data since the first snapshot.

The host then writes and additional 2GB of data which now brings the total to 12GB.

The Snapshot is then promoted and presented to another host, the host writes an additional 2GB of data in the space allocated for the TWS. Once the host is finished with the snapshot the TWS is then unmounted and removed. As shown in the image the total file system space needed was 22GB.

The following equation can be used here;

FS min =L + [A + ((N-1)*C] + [M*L]

 

Fully provisioned lun with virtually provisioned TWS

thick_thinktws

Now this configuration gives us the protection of a fully provisioned lun with a virtually provisioned TWS.

Once again, Celerra allocates enough space for a 100 percent data change (A) during the initial snapshot as a precaution. When the next snapshot is created, additional space is allocated equivalent to the amount of changed data since the first snapshot.

Because the TWS is virtually provisioned we reduce the total file system needed to just 14GB

The following equation can b used here;

FS min = L +[A + ((N-1)*C] + [M*T]

 

Virtually provisioned lun with virtually provisioned TWS

thin

Here Celerra takes full advantage of virtual provisioning and no longer is requires (L) published lun size and only requires (A) actual data.

The snapshot space does not require any initial allocation of (A) and the mounted snapshots require only the space for the changed data. The file system space needed to support this, just 6GB.

Here the following equation can be used;

FS min =A + [N*C] + [M*T]

If your wondering how to enable the sparseTWS support on the Celerra, check out my next post which covers this.

Also thanks to EMC’s Chris Stacey for hooking me up with the images used in this post.

Also if you have a EMC Powerlink account you can grab the sizing document from here.

Advertisements
Comments
  1. […] Posted in Celerra at 11:33 pm by Brian Norris My last post hopefully shed a little light on file system requirements for iSCSI luns on the EMC Celerra when replication or snapshots will be used. If you missed this and you think it might be of interest you can read it here. […]

  2. Chris Stacey says:

    Nice post. 🙂

  3. […] Long story short with a fully provisioned iSCSI lun, the minimum file system space required to perform a snapshot of the lun (and not taking into account changed data) is 2 x the published lun size, there are of course ways to reduce this required overhead and if you want to read more about this you can read one of my older posts about this here. […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s