Q: Difference Primary and Non-Primary Filegroup

[Simon_Eijs]

Hi,

The documentation / Help refer to Primary and Non-Primary filegroups for the Fingerprint DB:

The fingerprint database holds information about each Enterprise Vault single instance storage part (SIS part) that is stored in the group's vault stores. In addition to the primary filegroup, a fingerprint database has 32 non-primary filegroups, which hold the information about the SIS parts.

However, I don't understand what is exactly stored in the Primary and Non-Primary Filegroup. All documentation refer to 'SIS parts', for both Filegroups.

In addition, are there any guidelines on how much Storage must be reserved for Primary and (max 32) Non-Primary Filegroups?

• Or is it extended automatically?
• Does it depend on the amount of items (or sharable parts) being archived (we currently assume that e.g. we deal with around 300.000 journal reports per day (let alone the mailbox archiving)...

Thanks.

Simon Eijs

[Chris_Harrison]

The installing and configuring guide details the storage considerations on storage size, as does the performance guide which will be due another refresh prior to 8.0 SP1 shipping.

As for Primary/Non-primary filegroups....

The ‘Primary SQL File Group’ houses the system tables, stored procedures and views. The ‘SIS Data Locations’ house the actual fingerprint data for the Fingerprint Database, which is expected to grow very rapidly. 32 ‘Non-primary File Groups’ (each containing 1 file) are created, spread across the locations provided by the user as evenly as possible, and initially as small as possible. Therefore, a maximum of 32 locations can be entered in total (at 1 file group per location).

Whether it extends automatically or not depends on whether autogrowth is allowed or not.

Let me know if this helps.

[Simon_Eijs]

Chris,

Tnx. The thing I still miss is what's left in the respective archives? Let me give an example:

• User A sends an Email with a Word attachment to User B.
• The Email is Journaled.
  • If the Word attachment is found to be a 'Shareble' file:
    • Will the Word attachment be stored in the Journal Archive?
    • Will the Word attachment be stored in the non_primary FileGroup?
    • Will the remainder of the Email be stored in the Journal Archive?
• Mailbox archiving archived the Email of user B.
  • If the Word attachment is found to be a 'Shareble' file:
    • The Fingerprint DB will be updated.
    • The Word attachment will not be stored again.

If the Word attachment is stored in the Journal archive the first time a user request for 'Restore'or 'Read' must obtain the file from the Journal archive which is only supposed to be accessed by an Auditor... However, if it is stored in the non-Primary filegroup there is less of an issue...

I hope I explained my point.

Thanks,

Simon

[Simon_Eijs]

Sorry, but bullet lists fail in my final view. A simple rewrite...

User A sends an Email with a Word attachment to User B.

The Email is Journaled.

If the Word attachment is found to be a 'Shareble' file:

- Will the Word attachment be stored in the Journal Archive?

- Will the Word attachment be stored in the non_primary FileGroup?

- Will the remainder of the Email be stored in the Journal Archive?

Mailbox archiving archived the Email of user B.

If the Word attachment is found to be a 'Shareble' file:

- The Fingerprint DB will be updated.

- The Word attachment will not be stored again.

[Chris_Harrison]

>>User A sends an Email with a Word attachment to User B.
>>The Email is Journaled.
>>If the Word attachment is found to be a 'Shareble' file:
>>Will the Word attachment be stored in the Journal Archive?
If it is the first time the attachment has been archived, yes, it will be stored in the Journal Archive's partition.

>>Will the Word attachment be stored in the non_primary FileGroup?
The word attachment will be stored in the Journal Archive's partition. The RECORDs for the word attachment .(DVSSP and .DVSCC) will be stored in both the journal's vault store database, and the vault store group's fingerprint database's applicable MemberTable_XXX, on a non-primary file group.

>>Will the remainder of the Email be stored in the Journal Archive?
The per user information will be stored in a DVS file in the Journal Archive's partition along with associated journal's vaultstore database records.

>>Mailbox archiving archived the Email of user B.
>>If the Word attachment is found to be a 'Shareble' file:
>>The Fingerprint DB will be updated
Correct yes, the refcount in the applicable MemberTable_XXX will be updated. Additionally a record in the mailbox's vaultstore database will be create.

>>The Word attachment will not be stored again
Correct.

[GertjanA]

Hello Chris and Simon,

This is an very interesting albeit complicated part of EV8.0

I have an additional question for Chris, if you don't mind?

What happens if you have your Journal Store set to have an expiry of 6 months, and the Mail Store to retain items forever, with no deletion by the user allowed?

In other words, when the journaled email and attachment is being deleted by the expiry process, what happens with the SIS part? Does it stay in the Journal partition? Is it moved to the Mailbox partition? Is it removed, and is the reference in the fingerprint database updated to reflect it is now in the mailbox partition, and does EV then move the word attachment to another store?

I'm trying to get my head around this, but it might take a while for me to do so.

Thanks a mill!

Gertjan

[Chris_Harrison]

Sure no worries.

When a SISPart has been shared the refcount in the MemberTable_XXX is updated to reflect it is referenced in more than one item.

When expiry runs...

-The journal store's per user DVS file will be deleted immediately.

-The refcount to the word attachment is decremented. The actual DVSSP (and potentially DVSCC) will not be deleted or moved to the mailbox store.

-When the refcount reaches 0 will the on-disk files and the fingerprint records be deleted.

[GertjanA]

Hello Chris, thanks again.

Last question. Your answer indicates that (seeing our policy of users items are kept forever, no deletion allowed from mailboxarchive) attachments (items) stored in the journal partition who have references to mailboxarchives will never be removed from that partition. The effect might therefor be that the journal partition keeps on growing. I have read the insta;;/configure/planning guides, and am now thinking that we need to make two sharing boundaries. 1 for the journal servers, 1 for the mailbox-servers.

I also believe we might ask consulting for help prepping the upgrade.. !

Thanks again.

Gertjan

[Chris_Harrison]

Correct yes, the journal partition will continue to grow if SISParts are shared by the Mailbox vault stores (for when the mailbox tasks' archive the SISPart after the journal task has).

As for your vault store group boundaries...its a shame in your case as what you would really want to happen is to archive the data from the mailboxes first and the journal secon, that way data would be stored in the mailbox vault store and expiry would keep the journal vault store size down.

[Leonardo_Fern1]

The installing and configuring guide does not detail storage considerations. all it says on page 164 is "it is important to configure the fingerprint database appropriately for the amount of sharing in the vault store group". What is meant by "appropriately"? A system with 100,000 users needs more space than one with 3000. There are no guidelines. I have 3000 users. How much space should I allocate for my fingerprint database?

thank you