Duplication to a tape library attached to CASO

[EGarciaM]

I have a CASO Server (SrvCASO), a Managed Server (SrvManaged) and more managed servers planned but not installed yet. A tape library is going to be attached to SrvCASO and SrvManaged has a bunch of disks.

I'm trying to make backups in SrvManaged and link a stage that duplicates this backup as soon as it completes to the tape library attached to SrvCASO.

My tests so far have hit a wall as I'm not able to select the tape library on the duplication stage (it only shows "any storage" from storage pools but I cannot pick out one in particular). I'm testing with a disk storage at the moment (instead of the real library) but I assume this won't make a difference.

The reason in doing this is distributing backup jobs amongst a few managed servers freeing SrvCASO of this load and using it just for duplication jobs as it is the only one who has a library attached.

Is it possible?

[Gurvinder]

Need to have a disk storage or dedupe storage, depending on what kind of storage is configured on MBES. They need to be shared so that a stage can be added to duplicate the data to local disk storage on CAs and then it can be duplicated to tape on CAS from this local storage.

[Colin_Weaver]

You have four choices to get the data on tape (all of which require changes to configuration and some need further software licenses)

1) Move the tape library to the Managed Server

2) Connect the tape library to both servers (if they are in same location and SAN connectivity device sharing is available.)

2) Add disk storage to the CASO server and do the original backup to the CASO server (probaby only an option if what you intend to backup is on the same site)

3) Add disk storage to the CASO server and configure Deduplication on both servers with shared Deduplcuation folders, optimized duplication between the deduplication folders and then duplicate to tape stage. (often used where remote sites are involved)

[EGarciaM]

Thanks for your comments @Gurv and @Colin.

Option 4) seems the one that suits my needs. But when I try to go through with it I fail.

I define a job like this:

Storage is configured as:

And I get this error:

So, I create a backup from CAS, delegate it to MBES to do it to a MBES local disk, duplicate it to a shared dedup folder hosted on CAS and then try to duplicate it again to a CAS local disk (instead of the tape library that's not yet connected). And it fails.

I've tried the a job with the same stages but the last one and there's no problem creating the job.

So I assume the problem is with the last stage. But I don't understand why because the data is already "visible" by CAS after the first duplication. Shouldn't it be able to duplicate this data to a local disk? What am I doing wrong? Have I misunderstood you?

[Colin_Weaver]

Hmm well for starters you are not getting optimized duplication as the MMS does not itself own a Dedup folder. (Really the Backup Stage should go to Dedup on the manage server and then using 2 dedup folders means the data is not rehydrated between the servers.)

Because of this you have shared the CAS dedup folder so the Managed server runs the Backup and 1st Duplication stage wit CAS running only the 2nd duplicate

What I am not sure of is whether this needs MMS dedup shared to CAS so that the CAS runs both the 1st and 2nd duplicate stages. (with CAS owning non-shared deduplication folder and tape)

[EGarciaM]

Reviewing all the process I've noticed this that I've previously obviated:

On the backup stage storage screen it states that the server that you select applies to ALL backup jobs in this backup definition. So I guess it makes more sense of the error since if MBES is the server that's been used through all the stages then it won't be able to access to CAS local disk (of course). I wonder if I'm right... If so, then the only option is to duplicate to tape (CAS local disk in this lab environment) in another job definition, isn't it? And in this case, how do I link both jobs for the tape dup to begin as soon as the other job finishes?

To check this assumption I've created a dedup folder in MBES and have shared it with CAS. I've defined a job with a backup to MBES dedup, duplication to CAS dedup and finally duplication to CAS local disk (fake tape). If on the storage section of the backup stage I select CAS as BES the definition is OK, if instead I select MBES then the error appears. It makes very clear (at least for me) that what I want to accomplish (MBES backup + CAS duplication) in one single job definition is not possible...

[Gurvinder · Accepted solution]

do not share the dedupe that is created on the MBES. Create a backup definition which has backup going to this dedupe on MBES (no need to delegate it to this server). Since this dedupe is not shared, the job should automatically pick MBES server and run on this MBES dedupe. Now for the duplicate stage, you can select the CAS dedupe (which you shared with MBES). Once the duplicate stage is added, add another stage to duplicate from 1st Duplicate (dupicate set on dedupe CAS) to Tape attached on CAS (in this case the disk storage). This is the way you would achieve what you are looking for.

[EGarciaM]

@Gurv: You made my day, sir! After days of testing your solution runs smoothly!

Thank you very, very much!

One last question: I've done everything as indicated but I've inserted a "variation", the backup stage to a MBES local disk instead of to a not shared dedup folder, and everything works also. Is there any recommendation that tips the scale? I mean, which is better, a backup to a MBES local disk or to a MBES dedup folder.

[Gurvinder]

that depends on how much you will backup. With dedupe you can do client side where less data travel through the network to the mbes dedupe or when opt duping to cas dedupe only newer chunks are transferred (helps over wan. Initial seeding may be needed but post that opt dupe should be faster) but duplicating out to tape/disk on CAS would rehydrate the data so if dedupe ratio is high then rehydration may slow down the duplicate job rate. In addition to this how much retention do you need on disk storage. On dedupe you can have larger retention because data takes less space compared to disk storage. Dedupe demands more RAM than normal disk storage. So its important to evaluate these things before making up your mind

[EGarciaM]

Thanks again. You helped a lot.