cancel
Showing results for 
Search instead for 
Did you mean: 

What are the disk space requirements for the fingerprint database when configuring OSIS in EV 8.0.2

Leonardo_Fern1
Level 6
What guidelines can I use in allocating disk space for single instance storage fingerprint database? The Installing and Config guide just says I should allocate space appropriately. Well, what's appropriate for a 3000 user base?
1 ACCEPTED SOLUTION

Accepted Solutions

Wayne_Humphrey
Level 6
Partner Accredited Certified
Ok, so to work this out do the following:

Multiply the number of DVSSP “SISparts” files by 500 bytes to get the storage for the Fingerprint database.
Work out the number and size of DVSSP (SISPart) files 

Work out how many attachments we are archiving, and what size those attachments are. Typically, we’ll see around 20% of messages have attachments, and we’ll assume there is one attachment per message, and the average attachment size is around 200 to 250KB.  It’s useful to find out at this point from the customer what the predominant attachment types are.

Calculate the number of DVSSP files, as follows:

For simplicity we will assume that ALL attachments are over the shareable threshold of 20KB.  If you’re not comfortable with this (or the customer challenges it) simply use a slightly lower figure, say, 90%.  We are also going to assume that no message bodies are over the threshold.  In other words, we will assume that the DVSSP files are generated from only attachments.  In practice, there will be a little trade-off between the larger message bodies and smaller attachments… but let’s try to keep it simple!


Now, this is where we differ a little from mailbox archiving; we now need to divide the total number of journal messages by the “fan-out” factor to determine the number of original unique messages.  Assuming a typical average of three addresses per message, the fan-out factors are:

   1. For one journal mailbox, fan-out = 1.00
   2. For two journal mailboxes, fan-out = 1.75
   3. For three journal mailboxes, fan-out = 2.11
   4. For four journal mailboxes, fan-out = 2.31


Next, we need to factor in duplication of attachments which are attached to different messages. This is completely new to EV 8.0 and where we will really start to see some storage savings.  The challenge here is how to determine this “attachment-sharing” ratio.  For now, we should probably just use a “best guess” of between 1.5 and 3.0.  So, now take the figure you calculated in the previous step and divide by the attachment single-instancing ratio you have chosen.

And remember the Initial Storage Requirements.

Initial Storage required for the fingerprint databases (Clean Install)

A vault store group's fingerprint database holds the fingerprint, the storage location, and sharing boundary information for each SIS part that is stored in the

group's vault stores.

The fingerprint database has an initial storage requirement of 244 MB, made up as follows:

    *

      132 MB for the primary filegroup
    *

      1 MB for each of the 32 non-primary filegroups
    *

      80 MB for the transaction log device

The non-primary filegroups hold the SIS part fingerprints and other information about the SIS parts. If you share items using Enterprise Vault single instance storage, the non-primary filegroups may grow very rapidly in size. Ensure that there is adequate space for the non-primary filegroups to grow as data is added.

The New Vault Store Group wizard provides the following options for the initial configuration of the fingerprint database:

    *

      A default basic configuration, where Enterprise Vault locates the primary filegroup and all the non-primary filegroups on one device.
    *

      An advanced configuration option, where you can specify separate locations for the 32 non-primary SQL filegroups.

To ensure acceptable archiving and retrieval performance, it is important to configure the fingerprint database appropriately for the amount of sharing in the vault store group.

For optimal performance, do as follows:

    *

      Use the advanced configuration option to specify as many locations as possible on the SQL Server, up to the maximum of 32.
    *

      Use a separate device for each location. If you specify more than one location on the same device there is no performance benefit.

Note: To add or change locations after the fingerprint database is configured is a SQL Server administration task.

Limit transaction logs to an appropriate size for your back-up and maintenance plan.

--wayne

View solution in original post

2 REPLIES 2

jajensen
Level 4
Partner
As far as I know, there's still nothing out there that can provide any guidelines for the fingerprint database.  Best reason I can figure for this is because nobody can really say for certain just how much single-instancing would be done within any particular vault store group.

There are so many variables that it would be nearly impossible to say for certain how large that DB could get for X number of vault stores of Y items each.

I think the best advice to give would be to provide it as much space as you can, and keep a close eye on its growth.

Wayne_Humphrey
Level 6
Partner Accredited Certified
Ok, so to work this out do the following:

Multiply the number of DVSSP “SISparts” files by 500 bytes to get the storage for the Fingerprint database.
Work out the number and size of DVSSP (SISPart) files 

Work out how many attachments we are archiving, and what size those attachments are. Typically, we’ll see around 20% of messages have attachments, and we’ll assume there is one attachment per message, and the average attachment size is around 200 to 250KB.  It’s useful to find out at this point from the customer what the predominant attachment types are.

Calculate the number of DVSSP files, as follows:

For simplicity we will assume that ALL attachments are over the shareable threshold of 20KB.  If you’re not comfortable with this (or the customer challenges it) simply use a slightly lower figure, say, 90%.  We are also going to assume that no message bodies are over the threshold.  In other words, we will assume that the DVSSP files are generated from only attachments.  In practice, there will be a little trade-off between the larger message bodies and smaller attachments… but let’s try to keep it simple!


Now, this is where we differ a little from mailbox archiving; we now need to divide the total number of journal messages by the “fan-out” factor to determine the number of original unique messages.  Assuming a typical average of three addresses per message, the fan-out factors are:

   1. For one journal mailbox, fan-out = 1.00
   2. For two journal mailboxes, fan-out = 1.75
   3. For three journal mailboxes, fan-out = 2.11
   4. For four journal mailboxes, fan-out = 2.31


Next, we need to factor in duplication of attachments which are attached to different messages. This is completely new to EV 8.0 and where we will really start to see some storage savings.  The challenge here is how to determine this “attachment-sharing” ratio.  For now, we should probably just use a “best guess” of between 1.5 and 3.0.  So, now take the figure you calculated in the previous step and divide by the attachment single-instancing ratio you have chosen.

And remember the Initial Storage Requirements.

Initial Storage required for the fingerprint databases (Clean Install)

A vault store group's fingerprint database holds the fingerprint, the storage location, and sharing boundary information for each SIS part that is stored in the

group's vault stores.

The fingerprint database has an initial storage requirement of 244 MB, made up as follows:

    *

      132 MB for the primary filegroup
    *

      1 MB for each of the 32 non-primary filegroups
    *

      80 MB for the transaction log device

The non-primary filegroups hold the SIS part fingerprints and other information about the SIS parts. If you share items using Enterprise Vault single instance storage, the non-primary filegroups may grow very rapidly in size. Ensure that there is adequate space for the non-primary filegroups to grow as data is added.

The New Vault Store Group wizard provides the following options for the initial configuration of the fingerprint database:

    *

      A default basic configuration, where Enterprise Vault locates the primary filegroup and all the non-primary filegroups on one device.
    *

      An advanced configuration option, where you can specify separate locations for the 32 non-primary SQL filegroups.

To ensure acceptable archiving and retrieval performance, it is important to configure the fingerprint database appropriately for the amount of sharing in the vault store group.

For optimal performance, do as follows:

    *

      Use the advanced configuration option to specify as many locations as possible on the SQL Server, up to the maximum of 32.
    *

      Use a separate device for each location. If you specify more than one location on the same device there is no performance benefit.

Note: To add or change locations after the fingerprint database is configured is a SQL Server administration task.

Limit transaction logs to an appropriate size for your back-up and maintenance plan.

--wayne