No Catalyst Copy support?

[KarelV]

Hi Guys,

I am wondering if there is a possibility to copy backup files from 1 catalyst to another?
Previously we had 1 catalyst to store Both Datacenter 1 & Datacenter 2 backup copy jobs.
After a while the catalysts are running out of space.
We added a new StoreOnce catalyst so we can split up the data / jobs for each datacenter.

Thanks for the reply

[DGrinev]

Hey,

If I got it right, you are asking about the application driven feature called Catalyst Copy and it isn't supported by Veeam.
Please review this big thread for additional information. Thanks!

[foggy]

If you're looking for a one-time copy of files between Catalyst Stores, please review these options.

[JHuston]

Just wanted to tag onto this thread as another customer who needs Catalyst support for HPE Cloud Bank storage. This is currently keeping us from switching to StoreOnce appliances from our AltaVault.

[rdrost]

Foggy, do you have any information about the Catalyst Copy command line tool from HPE ? This is for evading the SOBR scenario when migrating individual veeam backup jobs to another Storeonce.

[foggy]

Yes, this tool is available through HPE support T4, so if you'd like to try it, you need to escalate the case with them.

[csydas]

Hey Foggy,

Can you elaborate on the magic words to get HPE to do this? We had a case open with HPE to do this and they insisted that no such tool exists, and trust me, I pushed and pushed for escalation until I was blue in the face, but to no avail. I was told the only tool they had was not safe for production and would not be usable in my situation. I even asked Veeam Support if they could figure out if there was a way they could put pressure on HPE, and they reported their higher tiers checked and no such tool exists per HPE.

Is there a specific person I need to ask for at HPE to handle this?

[NeilHPEStoreOnce]

Hi Csydas,

I hope I can help

I run the World Wide Product Management team for StoreOnce, and also manage the resale/relationship with Veeam.

Is there a CLI Catalyst Copy 'tool' or are support giving you the right answer that no such tool exists?
Both are technically correct.
There is code which we leverage in the HPE Catalyst plugins (the plugin for SQL for example) that performs a copy operation on Catalyst. L4 support has a version of that code without the 'application integration stuff' that just performs those copies which is used in some support cases to create copies of complete stores. Support are correct that the tool isn't generally used by support or available to customers; it's only used when L4 needs to and not GA as ultimately the unwary could do a lot of damage with it.

The issue is the tool is completely ignorant as to what is going on with any ISVs using the source Catalyst Store, so it would be entirely possible to corrupt your data by Veeam writing to an object while the tool is copying that object to your new store....as you can imagine....bad things happen.

For this reason the tool is not made available to customers via support and so support *is* giving you the right answer.

With Catalyst Copy integrated with Veeam (coming on the near term Veeam roadmap) that corruption can't happen as Veeam is in control of both writes and copying operations.

There are no magic words to get passed the support gate keepers, however, if you reach out to the team manager for WW StoreOnce PM we may be able to help you (neil.fleming@hpe.com)

Couple of things to bear in mind:

You'll need to suspend all operations to the appliance while the tool is in use.
The tool copies a full store as it has no way to map objects to particular Veeam jobs, so if you want to split your existing store you will need to:
- Copy the entire store to the 2nd StoreOnce using Catalyst Copy
- Have Veeam discover the jobs on the 2nd StoreOnce and delete what you don't want there
- Once you are happy with the 2nd StoreOnce condition, delete what you don't want on the 1st StoreOnce

Regards Neil

[csydas]

Neil,

Thank you so much for your post! I am very encouraged to hear that. I personally am on personal leave for the next few weeks, but I'll ask my colleague to reach out to you. I also hope that others in this thread stuck are able to reach out to you as well and get taken care of. I very much so appreciate you putting yourself out there publicly to assist.

[rdrost]

Well Foggy and Neil,
First , if the Calalyst Copy Tool can only copy whole stores at once it's not much or maybe even a worse scenario to use then the Veeam's SOBR method. I like to migratie invidual Veeam backup jobs including the backup chain. If you don't use a HPE Storeonce as a target this can be done in a simple way. Just copy out the files to new target and point backup job to it.

But i reached out to HPE also and after some effort got this explanation:

The Catalyst Copy tool won’t provide the goal you would like.
The way the tool itself is written is relying on a certain constant in a specific field called “Completed” in the Backup jobs, that is included in the Oracle RMAN, yet Veeam does not set this, thus the tool is unable to identify the Backups that needed to be actually copied over.

Although Development is working on the implementation, this needs to be completed with a close collaboration with Veeam, thus cannot be only completed on HPE side.

So please share your opinions !

[foggy]

I like to migratie invidual Veeam backup jobs including the backup chain.

How about using the workaround suggested by Neil above:

The tool copies a full store as it has no way to map objects to particular Veeam jobs, so if you want to split your existing store you will need to:
- Copy the entire store to the 2nd StoreOnce using Catalyst Copy
- Have Veeam discover the jobs on the 2nd StoreOnce and delete what you don't want there
- Once you are happy with the 2nd StoreOnce condition, delete what you don't want on the 1st StoreOnce

[rdrost]

True, if your usecase is that you have created one Store on your HPE Storeonce and you want to copy half of the backup jobs to a new Storeonce. Yes, then this method can help you to achieve that goal in the end. Veeam's SOBR method could not achieve that for you in my opinion.

In my usecase we have already devided the backupjobs over two Stores with more or less equally 30 TB in both Stores . So in my case i need only to migrate one Store and i am ready, all backup jobs can stay in the copied store on the new Storeonce. Veeam's SOBR would achieve the same in this case.
In both cases you cannot make any backups during the time of copy/evacuating 30 TB Data. This wil take a very long time
Above that we have also two stores for copyjobs of which one has to migrate to new Storeonce. These Stores contain a lot more data (Archived Monthly).

So if one could migrate jobs one by one this would be much more convenient and with less risk.

[foggy]

I agree, but currently this is not possible.

[NeilHPEStoreOnce]

Hi Csydas & Rdrost,

Csydas - yes happy for your colleague or anyone else on the Veeam Forums to reach out to me on any matter relating to StoreOnce (or the wider HPE).

Rdrost - The tool can copy individual catalyst objects. So when it is implemented in say the catalyst plugins for SQL/Oracle etc, it knows when an Oracle job finishes (that completed flag you mention) and it knows what objects relate to what Oracle backup job because the code created those objects and has a catalog relating them. The 'headless' code used by L4 can still copy individual catalyst objects but wasn't involved in the creating of the objects. There is no way for the user (or HPE) to relate individual objects on the catalyst store to a particular Veeam backup (only Veeam knows which object make up with job), so for a user to try to replicate individual objects in this manner would almost certainly end in disaster.

That's why in my example to Csydas I suggest replicating the entire store (when it is in a state where everything is finished), then letting Veeam delete what isn't needed; Veeam, and only Veeam, knows what isn't required for the jobs to be retained on the 2nd StoreOnce.

What you want is Veeam Catalyst Copy - but unfortunately that is coming until later from Veeam.

Regards Neil

[rdrost]

Thanks for your input.
But know stepstoning to an earlier conversation in this thread about "dedup" method of the upcoming archiving support for S3 storage in Veeam 9.5 update 4. But it is surely related for my usecase in finding solutions for expanding our Veeam Storage capacity the most efficient way.

On Mirazon.com theres a statement about Archive Tier 2.0 . In states that you can have your local backup jobs coninue to do a 7-10 days cycle on fast storage while also storing it to your achive tier in the same backup yob. Before you had to set up a backup copy job alongside.
And now its done in one step. Without replacing backup copy jobs it might allow your backup jobs to have a more traditional backup structure in keeping long-term restore-points.
Well this may well be true but in my opinion you still cannot do without Copy Jobs if you still want to hold on to a GFS schedule for long-term archiving.
And if you hold on to GFS archiving then there's at least one extra read step . If your Copy Job lands on Fast Storage it puts extra burden on your fast storage.

If you choose to use only backup jobs for keeping your for example 7 daily , 4 Weekly Full end 12 Monthly Full you might benefit by not using backup copy jobs. In the above scenario one less read and write to make on your primary fast storage. But Archive Tier 2.0 does add one read again on your primary fast storage. But secondly you have to create more backupjobs for the same VM's. I think one for the Daily, one for the Weekly and One for the Monthly. And all together theres exta burden on your primary storage.


But since Veeam 9.5 Update 4 is not generally available yet i might see this compltely wrong. So Veeam, please share you opinion.