Video Screencast Help
Symantec Appoints Michael A. Brown CEO. Learn more.

EV10 Tiered Storage questions

Created: 07 Jun 2013 • Updated: 21 Jun 2013 | 3 comments
This issue has been solved. See solution.

We are looking to deploy EV10 within the next few months, but I'm trying to get a better understanding of how the storage works. 

We want to deploy a tiered storage model, using fast storage for indexes and new data, with mail migrating to slower storage after a period, and eventually migrating to tape.

On this basis: 

- In a multi-server environment, how does the server interact with the storage layer? Are there any concerns with file locks etc if one of the servers goes down?

- How does mail move between tiers? What service is responsible for this? What happens if this service fails?

- How does SIS affect this? Is it possible for a new piece of mail to match one that has already moved to tape? If so, how will EV address this?

Any documentation/process diagrams which can provide some insight on the above would be much appreciated. 

Thanks

Operating Systems:

Comments 3 CommentsJump to latest comment

JesusWept3's picture

- In a multi-server environment, how does the server interact with the storage layer? Are there any concerns with file locks etc if one of the servers goes down?

EV storage servers are currently 1:1 but the OSIS model means that a single server going down could affect many archived email because of being shared

- How does mail move between tiers? What service is responsible for this? What happens if this service fails?

Enterprise Vault collects the items into CAB files and after a certain age is then pushed to the secondary storage.

This is all governed by the Storage Service and handled by one of the storage processes

For instance collections is handled by StorageFileWatch, retrieving items from migrated cab files handled by StorageOfflineOpns

If the processes fail then items won't migrate or be retrievable I suppose, just monitor the event logs and space used I'm tertiary storage to ensure items are being moved etc

- How does SIS affect this? Is it possible for a new piece of mail to match one that has already moved to tape? If so, how will EV address this?

EV hashes each item over a certain size that can be SIS'd, if an item gets archived and the hash matches, it simply creates a a database entry linking it to the existing file and does not create a new file

Also lastly I would highly recommend against tape migrations, I don't think there is a single positive side effect from it and when you have to rebuild an index that takes weeks because of tape swapping, you'll have wish you hadn't

SOLUTION
GertjanA's picture

Hello TImbstoke,

First, if you have budget, involve a partner. We (partners) have access to several tools that help in determining the best setup and configuration for a new EV installation. I personally think any new installation should be done with help from a partner, because we know what to do in regards to preparation, configuration etc. I consider EV not to be an 'install out of the box, off you go" product. It requires understanding, and good management buy-in!

As for your questions:

- In a multi-server environment, how does the server interact with the storage layer? Are there any concerns with file locks etc if one of the servers goes down?

If one of the servers goes down (assuming you mean an EV-server) this will affect users handled by that EV-server. They cannot use EV anymore. If you have DR using building blocks configured, that is relatively easy to resolve. If a file-server (ie the storage itself) goes down, that will also stop EV. There are build in checks to prevent EV doing something when important parts are not available.

- How does mail move between tiers? What service is responsible for this? What happens if this service fails?

You should read up on 'collect and migrate'. What you configure is the following (EXAMPLE!!!):

On a Vault Store Partition, you enable collections to collect archived data in CAB files. Collect data older than 30 days. Configure Migrate to migrate those CAB files to secondary storage after 30 days, and remove the original cabfiles after 1 week. This effectively means you keep archived data on the EV-server firstline storage for 2 months and a week. Migrating from secondary to tape I am not too sure about, but that is something a partner should be able to explain. This is done by the storage-service afaik. If the service fails, firstly no items are being stored, and you see the MSMQ Storage going up. Usually a restart of this service fixes this. Also check https://www-secure.symantec.com/connect/articles/how-collections-and-sparse-collections-work for an explanation. Alsolook on the forum for 'collection', you'll find loads of entries there. A starter might be http://www.symantec.com/docs/HOWTO57117

- How does SIS affect this? Is it possible for a new piece of mail to match one that has already moved to tape? If so, how will EV address this?

If a new item is archived, and an attachment is already 'somewhere' in EV, the item will NOT be stored, but a reference to the item will be made in SQL. If the new item is needed by the user, he retrieves it, and EV will get it from wherever the original is stored. SIS is based on fingerprinting, and helps a lot on limiting necessary storage. This also goes if you want to archive File servers, or Sharepoint, or even with importing PST's..

I am not sure if proces diagrams are publically available, but a partner should have those too.

Also check this for video's. Some are very common, some are very good: http://www.symantec.com/docs/TECH68577

 

Thank you, Gertjan, MCSE, MCITP,MCTS, SCS, STS
Company: www.t2.nl

www.quadrotech-it.com

www.symantec.com/vision

TheEmptyMind's picture

- In a multi-server environment, how does the server interact with the storage layer? Are there any concerns with file locks etc if one of the servers goes down?

If you're going to run EV on a cluster (which is a supported scenario), the resource failover and failback on EV nodes is akin to SQL/Exchange failover/failback.  A blip is inevitable but is hardly noticeable.  We use non-clustered enviornment for EV and use building blocks to failover/failback between servers.  See http://www.symantec.com/business/support/index?page=content&id=HOWTO37984 for more on building blocks.

- How does mail move between tiers? What service is responsible for this? What happens if this service fails?

You will need to enable Collections on the vault store and configure "Migration" to specify your criteria on when (i.e., age) to move the files.  See http://www.symantec.com/business/support/index?page=content&id=TECH71338.  NetBackup and EV have tight integration that lets you use tiered storage -- see http://www.symantec.com/business/support/resources/sites/BUSINESS/content/live/TECHNICAL_SOLUTION/127000/TECH127059/en_US/340125.pdf.

- How does SIS affect this? Is it possible for a new piece of mail to match one that has already moved to tape? If so, how will EV address this?

SIS is unaffected (see http://www.symantec.com/business/support/resources...) but you will need to have an online tape storage if you let users retrieve old items.

From the doc:

"Since items are not changed once they are written to disk, it is now possible in Enterprise Vault
8.0 to collect archived items much sooner than was possible before. Collecting after 1 day for
example would be perfectly reasonable as this would not have any affect on any further potential
sharing of those items".
 
Hope this helps!