Ptotection Manager - Dataset

[sheelnidhig]

I have a questions regarding Creating and managing Datasets in Protection manager:

1. Do we have any guidelines when creating datasets, situation is:

     * We have few Hundreds of filers and Windows server, taking backup using snapvault and OSSV.

     * Planning to put things on Protection manager for data protection.

** How we shall name the Datasets, so that it is easy to identify the resources within it ?

** Do we have any common interface which can list resources within the Dataset, something like a tree view or dataset explorer or anything which can do the job.

The Complication here is we cannot create individual dataset for each windows server or each Volume from the Filer.

Reason is each dataset will use one destination and that cannot be used for other dataset.

It looks to me like a one to one relation, one destination volume can protect only dataset.

Please suggest the best possible way to handle this situation,

Thanks,

Sheel

[mschlaerth]

Hello Sheel,

My strong recommendation for creation of datasets is to:

1.  Group what you plan on putting in  them based on the policy you are  going to apply to the set... i.e. if you are talking specifically about OSSV, consider the frequency of the updates, and how long you are going to keep the backups.  that schedule and retention pretty much define the policy.  if you have 2 or 3 variations regarding SLA's. then that's the first logical dataset divider.

2. Group based upon network load balancing.  Once you ahve established SLA driven differences that provide a list of what would go into the dataset, then you can make a determination based on scheduled updates.  If you have 250 OSSV hosts, 100 in dataset A, 100 in dataset B, and 50 in dataset C, you may want to break dataset A into smaller datasets, and offest the scheduled update times so that it balances the network load, timing them so that as one update completes, the next is starting up.

3. You can also group based on destination...  It's not a true one dataset to one destination relationship scenario.  a dataset can have dozend of VSM relationships in it, or it can have a single snapvault.  Provisioning manager prefers to create a single volume per OSSV dataset, and then creates qtrees under it for the actual backup paths on the hosts.  Filer snapvault creates more of a volume to volume ratio, then makes qtrees within on the destination.  Based on that, and whether you have 8.x on your destination, you might break up the OSSV jobs around how big you anticipate them getting, or creating a single volume oin a 64bit aggregate, and pointing all of your datasets manually at it as a secondary.

Clear as mud.....