Possible to determine Single Instance Storage ratio of Enterprise Vault on Centera storage

[AaronJAnderson]

We are curious about the effectiveness of SIS (single instance storage) on our system. This seems like it is a hard statistic to retrieve from systems. Our Centera will give us SIS reports, but it is essentially zero because of Enterprise Vault's advanced SIS. Is it possible to get this stat from EV? 

[JesusWept3]

Enterprise Vault 8's OSIS for Centera does not apply due to the fact that the Centera does its own SIS (some may call it de-duplication), so the Centera has always been able to share out the attachments and such

[AaronJAnderson]

So at which level is SIS happening?

Is EV doing SIS, or is our Centera doing SIS? We ran some kind of report on the Centera and it didn't show any gains for deduped stuff... which doesn't make since so I assumed that EV was doing it at a software level insteaad of hardware.

[AaronJAnderson]

Anyone?

[Bruce_Crankshaw]

I have several customer using Centera and I investigated the question around EV/Centera SIS ratio
My findings were as follows
1. As mentioned Centera is 100 % responsible for SIS .Make sure you do tell EV on the partition level "To Enable Device Level Sharing
2.For me the frustration came from how do I now tell my customers with conviction  they are getting SIS on the Centera.This is really a Centera deliverable ...the rule being below .I took this from the EV\Centera best practice whitepaper .In summary we have to rely on the Centera that is working correctly around how it utilizes SIS ...well at least that was my understanding :)

 wha

[Liam_Finn1]

I agree with Bruce. We have done extensive work with EMC and have asked many many  questions on this.

In most cases where a Centera is used with Archiving the Centera is set to Performance.

In Performance mode any item less than 256k is not single instanced because it would take too much processing time. Any item greater than 256K would be single instanced. To get numbers on the SIS it would require examining the size of each item after compression to see if it meets the 256K marker then enumerate the totals of all items less that 256K and add them. Then figure out what items over 256K are archived, then figure out the number of times each was archived and then sis that to one copy of each on centera against X number showing in EV

To work this out I hope you have lots and lots of time free to make all these calculations and if you find a way to automate it call me because I sure can use it. I have given up trying to figure it out because life is too short to try to find the end of a bottomless pit
 

[AaronJAnderson]

Good information guys. Have you ever seen this? Mine isn't working, of course, but it could be a good start.

[Bruce_Crankshaw]

Hi Aaron
Yes that would be useful but once again it won't apply to  understanding Centera  SIS
Also I suspect the reason that report is not working is because your EV SQL reporting isn't't working
Aaron send me a PM with your email address and I'll mail you the EV/Centera Best Practice White paper


@Scanner
lol...your comment "because life is too short to try to find the end of a bottomless pit" is exactly how I feel about  Centera\SIS ,,,even though it is interesting and relevant .There are far too many things about EV I can quantify and logically prove ..I would rather spend my time on those :)

 

[AaronJAnderson]

I had the reporting services installed on another server. I'm sure it's just looking at the localhost for that report. PM with my email coming soon.

[Chuck_Ford]

If anyone knows of a way for Centera/EV to report on the SIS performance on Centera that would be very, very useful.  With a large amount of email stored on Centera for regulatory purposes (in Compliance Plus mode), we have decided to continue to use the same Centera nodes for our Mailbox Archiving rollout... with the expectation that it would handle the de-duplication of attachments.. but we need reporting to show that the higher cost Centera is justified due to this de-duplication.  Granted, if we used native EV8.x SIS with a cheaper NAS storage solution, we'd also get similar (I assume) SIS benefit, but we'd then also have the additional overhead of another storage solution, the SIS database (not yet internally certified), and the backups of that data (which we don't currently need due to the Centera replication solution in place).

[Chuck_Ford]

Does anyone have the link to the Centera BP guide... no luck searching the support site.

[Liam_Finn1]

Chuck,

Depending on the number of users you have Centera may not always be the answer. Centera is not the fastest way of storing data. If you will have many users that will access their archives daily a SAN may be the better solution.

[Chuck_Ford]

Yes, Centera is not the quickest for retrieval but we've implemented Virtual Vault only (no mailbox shortcuts) with Vault Cache (and Outlook Cache Mode) so the users will have limited need to "see" that 1-2 second delay.  The reporting data for de-duplication would be useful for the Regulatory side either way.

[Liam_Finn1 · Accepted solution]

I have been trying to find a way for some time but no luck


What I do is i do a comparisson on what EV says it has archived against the numbers from Centera (capacity only)

You cant do it by C-Clip because each C-Clip can contain a number of BLOBS
You cant do it by BLOB because any large items (greater than 50MB)create two BLOBS for each item

Each of these will skew the returned results so I recommend giving up on doing an Item level report. Best option is to stick with SIS on the capacity level because that you can gather by using show capacity on the centera and usage.asp on EV.

Depending on if you are using CPM or CPP on the Centera it will vary how you address the results found