cancel
Showing results for 
Search instead for 
Did you mean: 

Keeping the past 3 years archived data on Symantec EV by migrating the rest of closed partition to tape using Netbackup ?

John_Santana
Level 6

People,

I'm planning to reduce the disk space usage of the EV 9.0.4 LUN by sending all of the data older than 3 years off to tape by using Symantec Netbackup v7.6.

How can I achieve this while still allowing user to retrieve the archived email & file when they double click on the Outlook archived email and also the file server place holder ?

Once I migrate the existing EV 9 to EV 10, can user still receive those archived item written to tape after the EV 9 server & SQL 2005 database has been decommisioned ?

1 ACCEPTED SOLUTION

Accepted Solutions

Merv
Level 6
Partner

Hi John,

Yes your plan is correct however please note the timelines  can be longer than expected and defer from env to env.

What is the storage team or mgmt plan for the old storage provisioned to the current EV 9.Are they planning to decommission it?If  they are not going to do that, they can simply assign the old storage to the new server. The EV server can support multiple storage types i.e. CIFS and NTFS concurrently.i.e. old partitions are on NTFS and new ones are on CIFS.

Seen lots of orgnazation leave out storage expiry plannning or scheduling so just sharing my thoughts on this

Storage expiry is used to really get rid of really old data which is no longer required - i.e. as a business policy and end user needs. However as a secondary effect it does lower Storage TCO(h/w running cost inc support/licensing/etc). And if that is the plan you would need to weigh the pros and cons. Lower TCO vs Users not getting access/  easy access to old archived data.  The concept of tiered storage  can also comes in..i.e. moving old archives to cheaper/possibly slower/lower HA /reduced functionallity(offline/tape)

As a side note am expanding on retention policies and managing user expectations for access to old email. Also remember to  communicate it the retention policies/periods to the users.

1. Set up Journal retention policy periods  to comply with the business needs - the usual suspects are Legal, Compliance,HR, Financial folks who have thier own laws /best practices governing these like SOX, HIPA etc which may be like 7 years or 5 years. After everyone has agreed what is the longest time you need emails for, you can set the policy, size and grow your storage gradually.  Monitor this on a regular basis and as well as the storage reductions when expiry starts running.

2. User mailbox retention policies - This does not need to be the same as the Journaling ones.  You can set this to a lower retention period.  EV users are infamous for not deleting mail  that they have an unlimited quota for archived mail - they may get paranoid when storage expiry comes along. The alternative is to have  users have a way to retain certain important reference mails for a longer period - they can have special retention folders or have this mails kept in a shared mailbox with longer retention periods assigned.All that said and done you are not really going to  "save" space if OSIS is setup and if you have a fingerprint DB. You can confirm this from the EV usage reports - most of the emails here will just be DVS pointers back to the Journal Vault Store. 

 

Hope this helps - good luck

View solution in original post

12 REPLIES 12

JesusWept3
Level 6
Partner Accredited Certified

OK if it were me, and i absolutely HAD to migrate to tape, i would do it AFTER you do your upgrades.
Because when you go to 10 or 11, its going to retrieve all those emails for things like converting to 64 bit indexes, or building the MetaDataStore for the Fast Browsing in EV11 (needed for IMAP and recommended for the new search)

You can read more about native migration here:
http://www.symantec.com/business/support/index?page=content&id=TECH49714

Honestly though, I couldn't recommend it, its painfully slow, and if you have to rebuild indexes or decide to roll things out like imap or vault cache that demand those items , it can slow down processes by quite a large margin

https://www.linkedin.com/in/alex-allen-turl-07370146

AndrewB
Moderator
Moderator
Partner    VIP    Accredited

How can I achieve this while still allowing user to retrieve the archived email & file when they double click on the Outlook archived email and also the file server place holder ?

It's built-in to the mechanism. Just like with collections (cab files) where you change the location of the archived items, EV updates the vault store databases accordingly so when a user retrieves an archived item the back-end takes care of the rest.

 

Once I migrate the existing EV 9 to EV 10, can user still receive those archived item written to tape after the EV 9 server & SQL 2005 database has been decommisioned ?

If by migrate you mean server/OS migration then i dont see why there would be any difference. After all, you will still have your same databases and EV config - just on new servers.

AndrewB
Moderator
Moderator
Partner    VIP    Accredited

..and like JW said, i would strongly encourage against it for the same reasons he listed. both collections and migration to tape have a lot of disadvantages, some of which JW listed for you.

John_Santana
Level 6

many thanks for the clarification Jesus,

The problem that I'm curently facing is that I do not have enough space with the same amount of Existing disp space to provision on the SAN to be presented to the new EV server.

Existing Configuration:

PRODVAULT1 - EV 9.0.4 on Windows Server 2003 32 bit
C:\ - 50 GB
D:\ Index 1 TB
E:\ VaultStorepartition1 (Closed) 1 TB
....
W:\ VaultStorepartition19 (Closed) 1 TB
X:\ VaultStorepartition20 (Opened) 1 TB 


PRODSQLDB13 - SQL Server 2005 SpP4

New Configuration:

PRODVAULT2 - EV 10.0.5 on Windows Server 2012 R2 Std.
C:\ - 80 GB
D:\ Index 1 TB --> I can provision this as block storage to the new EV 10 VM.
\\CIFS-Srv01\EV\VaultStorepartition20 -- (Opened) 1 TB --> --> I can provision this as CIFS share to the new EV 10 VM on EM C data domain


PRODSQLDB17 - SQL Server 2012 SP1 Std.

Do I have to provision the other closed partition LUN as well in the new EV 10 server ?

 

 

John_Santana
Level 6

Andrew, so what's the suggested migration steps or approach that is generally used by the Symantec EV customer ?

Merv
Level 6
Partner

John,

I have had some exp with the feature and the way storage cost are going down by the min it does not make any sense whatsoever in your scenario. I'll go into that in a minute.

 

1.have you read http://www.symantec.com/business/support/index?page=content&id=HOWTO84125

and especially the first page? Ok for benefit of getting this message across quick here it is..

----------------

Considerations before Setting up Migration

What Types of Data Should Be Migrated?

It is highly recommended that the NetBackup Enterprise Vault migrator not be used in conjunction with email related archives (especially when migrating to tape) due to the potential sheer volume of archived content. If the Enterprise Vault environment will be performing numerous recalls of archived emails (such as using EV Browser Search, Discovery Accelerator, Clearwell, exporting archived emails to PST/NSF, or verifying Vault Store consistency using the Enterprise Vault evsvr utility), numerous restore jobs will be created on the NetBackup Master server. Index rebuilds can also cause a mass recall of content from NetBackup.

If the tapes that contain migrated Enterprise Vault content are not available in the tape library, the recall of the archived content will be further delayed. Lastly, if the NetBackup environment is busy with a limited number of tape drives, restores can also be delayed as restore jobs generally have a lower priority for tape and disk resources compared to that of scheduled backup jobs.

If Not yet at Enterprise Vault 10

If the environment will be upgraded to Enterprise Vault 10, it is recommended that migrations be configured after the upgrade has been performed and if 32-bit indexes will be upgraded to 64-bit. If content has been migrated before indexes have been upgraded, content will be recalled back from NetBackup during the index rebuild process. This process can be very slow if there has been a large amount of content migrated to NetBackup. It should be noted that it is not immediately necessary to upgrade 32-bit indexes to 64-bit after an upgrade to Enterprise Vault 10. 

----------------

There are very specific use cases for the feature...very very old data which does not need to be accessed by the average joe .

To go on a short rant explaining the above in my own words

- does not need to be search and RETRIEVED  say by clearwell or the accelerator products.

-Shortcut expiry has been run on them so no one is going to accidentally recall them or have their offline vault start initialising...

-indexes? You don't upgrade rebuild any index volumes or that they never get corrupted by any A/V scanners or any disk corruption in any form or shape. Index files are especially sensitive to any poking around.

-You don't have any DR scenarios where you los data causing you have to do a data integrity checks with the EVSVR tool...

-exports to pst...

-any recalls in any way shape or form will cause NBU restore jobs to start requesting 20mb-99mb cab flies from a tape somewhere in some library hopefully still mounted and not sent offsite or worse expired or recycled or corrupted. Have I mentioned archcabs?

 

anyway all you need is to plan your server cutover and mount those volumes so that the new server "sees" your old (NTFS on VMFS) and new storage (DD). There is a tech note on partition moves in case drive letters or unc paths change. Basically updating your enterprise vault dir db partition enter table..easy..it will save you all the considerations...

John_Santana
Level 6

ok, here's the compiled steps that I have finally collected after browsing around the Symantec connect forum to do the migration, please let me know if there is anything that is missing or needs to be changed:

Data purging & PST exports

These steps must be done to reduce the total disk space usage by EV in overall by keeping the past 2 years data only. Exporting to PST will make the restore relatively easier later day after the old EV 9 is decommissioned and the data is purged by EV before upgraded into EV 10.

  1. Backup the existing EV databases into temporary disk location or to tape.

http://www.symantec.com/business/support/index?page=content&id=TECH35744

http://www.symantec.com/connect/videos/how-move-enterprise-vault-sql-databases-another-sql-server

  1. Export all the items older than 2 years from all the archives to PST files.  This may take very long time depending on the size of archives.
  2. Change retention setting to 2 years and wait for deletion process to complete.

 

Move / Migrate the existing VaultStore to EMC Data Domain CIFS share\

This step must be done before migrating to EV 10 to make the CIFS share readily available from EV 9 into EV 10 server.

  1. Move Vault Store partitions and Index Volumes to new storage

http://www.symantec.com/docs/TECH35742 - for Vault Store Partitions

http://www.symantec.com/connect/videos/enterprise-vault-9-how-move-index-volume - For index volumes

 

Preparing the new EV 10 server

Once the server has been provisioned with the CIFS share presented. Use EV migration wizard http://www.youtube.com/watch?v=tQh4ZfIn99A and http://www.symantec.com/docs/HOWTO42445 

1.Put EV into backup mode

2.Copy all the data (Index Location/Vault Store Partitions) to the new server with the same letters.

3.In the new EV 10 server, Run Migration tool to create a package

4.Update EV server alias to point to new server

5.Run Migration tool and load package created above

6.Go through Wizard and check the status

Merv
Level 6
Partner
Hi John, I still dont get what your objective is. Are you doing a migration AND a storage deletion? "reduce the total disk space usage by EV in overall by keeping the past 2 years data only. Exporting to PST will make the restore relatively easier later day after the old EV 9 is decommissioned and the data is purged by EV before upgraded into EV 10." Is the client specifically looking to reduce storage? The best practice is to use storage expiry...exporting everything out to PST as a method to retain data is slow and painful and you will have PST sprawl and getting data out of PST is a pain if you do need it. PST still take space (Fyi- EV was specifically designed to avoid the nihhtmare of having to manage hundreds of 20gb pst) You are just better off doing a good VM snapshot of the entire system and db (if VM) or full backups and keep those somewhere safe in case you need the data for DR . Then turn on scheduled storage expiry either after you move to 10 or before...as per business needs i.e. if current retention is forever or 7 years or 10 years than you reduce it to save on storage.

John_Santana
Level 6

Merv,

Thanks for the reply. Yes I'd like to do Data Deletion, and then followed by Storage Migration from NTFS LUN into CIFS share and then after that followed by the EV 10 Upgrade if that is the correct stage ?

Yes, reducing the disk space is needed here because there is not enough Storage Array capacity to provision new EV LUN with the same size as the old EV server.

Unfortunately the EV Virtual Machine already got 18+ NTFS Phyisical RDM so taking snapshot is not possible. The SQL Server 2005 Database is physical old box, hence the needs to upgrade to EV 10 due to the compatibility of EV 9 with SQL Server 2012.

Merv
Level 6
Partner

Hi John,

Yes your plan is correct however please note the timelines  can be longer than expected and defer from env to env.

What is the storage team or mgmt plan for the old storage provisioned to the current EV 9.Are they planning to decommission it?If  they are not going to do that, they can simply assign the old storage to the new server. The EV server can support multiple storage types i.e. CIFS and NTFS concurrently.i.e. old partitions are on NTFS and new ones are on CIFS.

Seen lots of orgnazation leave out storage expiry plannning or scheduling so just sharing my thoughts on this

Storage expiry is used to really get rid of really old data which is no longer required - i.e. as a business policy and end user needs. However as a secondary effect it does lower Storage TCO(h/w running cost inc support/licensing/etc). And if that is the plan you would need to weigh the pros and cons. Lower TCO vs Users not getting access/  easy access to old archived data.  The concept of tiered storage  can also comes in..i.e. moving old archives to cheaper/possibly slower/lower HA /reduced functionallity(offline/tape)

As a side note am expanding on retention policies and managing user expectations for access to old email. Also remember to  communicate it the retention policies/periods to the users.

1. Set up Journal retention policy periods  to comply with the business needs - the usual suspects are Legal, Compliance,HR, Financial folks who have thier own laws /best practices governing these like SOX, HIPA etc which may be like 7 years or 5 years. After everyone has agreed what is the longest time you need emails for, you can set the policy, size and grow your storage gradually.  Monitor this on a regular basis and as well as the storage reductions when expiry starts running.

2. User mailbox retention policies - This does not need to be the same as the Journaling ones.  You can set this to a lower retention period.  EV users are infamous for not deleting mail  that they have an unlimited quota for archived mail - they may get paranoid when storage expiry comes along. The alternative is to have  users have a way to retain certain important reference mails for a longer period - they can have special retention folders or have this mails kept in a shared mailbox with longer retention periods assigned.All that said and done you are not really going to  "save" space if OSIS is setup and if you have a fingerprint DB. You can confirm this from the EV usage reports - most of the emails here will just be DVS pointers back to the Journal Vault Store. 

 

Hope this helps - good luck

John_Santana
Level 6

many thanks for the detailed explanation Merv.

As at the moment I'm not implementing the DA/CA feature because I;m not sure how it is working ? or if it is consuming more disk space / separate server needs ?

What is OSIS and Fingerprint DB ? in my SQl Server I've never seen Fingerprint DB.

Merv
Level 6
Partner

ok when you say that you are not implementing DA/CA do you mean you don't have exchange Journal archiving. Then my scenario will not apply

OSIS -Optimized Single Instance Storage (OSIS)

This comes with a Figerprint DB or Vault Store Group DB. - allows you to share across vault store DB's