Wan Acceleration Global Cache copy

[Butha]

Hi All,

I saw in a video with Gostev presenting a mention of a "hack" to run a backup copy job first to local storage using wan accelerators- then copying the global cache folder to target side (Accross a WAN) and then creating a new backup copy job specifying the target wan accelerator proxy - which would greatly reduce the "first run". I am aware that exporting the source VM to removable storage, transporting to target and either using it for seed, or restoring the complete VM to target environment and using that as a map target to will also generate the "global cache" on the target side, but this approache still takes a long time for big VMs. (500GB +)

Can anybody comment on this "hack" ? (words used in the presentation).

Questions I have should this be possible:

What folders do you need to copy in the global cache? (there are a number of folders, as well as a root folder). There is also a sub folder with the source VM - perhaps copying this alone to target global cache folder (with other cache data in there already from other jobs) would be correct?

I'm busy testing to achieve this myself, but would be great if anybody else has been able to do this.

Butha

[foggy]

Butha, to seed the WAN global cache, you can copy the GlobalCache\trg folder to the target WAN accelerator machine (preserving the path) and restart the Veeam WAN Accelerator Service.

[Butha]

Thank you! I will give it a go.

B

[xianmacx]

I am having trouble finding the link to the "hack". I am also trying to find a way to seed the global cache.

I recently seeded a backup copy, but it still took 100 hours to build the remote cache on first run.

Thanks,
Ian

[Vitaliy S.]

Ian, have you copied the entire cache folder as foggy has suggested above or you've used existing backup files to pre-seed your backup cop job?

[xianmacx]

Vitaliy,

Steps I have been taking...

1. At the on-premise site, backup jobs run to the nas.
2. At the on-premise site, I create a backup-copy job and just save it to the same nas. Since its all done at one site, I use "direct" instead of picking WAN accel, which as far as I know doesn't create a cache?
3. Take that backup-copy file that I just created, move it via usb to my datacenter.
4. Map the backup copy job to these newly transferred files at the datacenter. This works, it just takes numerous hours on that first run as there is no cache built. I need to be able to seed that cache, not only the "backup-copy" job files.

Thanks again,
Ian

[foggy]

I am having trouble finding the link to the "hack". I am also trying to find a way to seed the global cache.

Seeding WAN cache procedure is described in my first post above (copying the GlobalCache\trg folder). Have you enabled WAN acceleration on the backup copy job?

Steps I have been taking...

Have you completed all the steps outlined here while seeding your backup copy job?

[veremin]

2. At the on-premise site, I create a backup-copy job and just save it to the same nas. Since its all done at one site, I use "direct" instead of picking WAN accel, which as far as I know doesn't create a cache?

Yep, the direct copy job doesn’t have any cache, at all. So, if you wanted to seed Global cache, you should have run local Wan accelerated job, instead. Thanks.

[xianmacx]

Vladimir,

So instead of picking "direct" when I do the initial "backup copy", I can just select the local accelerator for both source and target and then run the job?

For some reason I assumed it would not let me pick the same wan accelerator for both source and target...

[foggy]

So instead of picking "direct" when I do the initial "backup copy", I can just select the local accelerator for both source and target and then run the job?

Correct.

[xianmacx]

Great, I will give that a try.

Thanks for all the help.
Ian

[foggy]

So instead of picking "direct" when I do the initial "backup copy", I can just select the local accelerator for both source and target and then run the job?

My bad, I missed the fact that you were talking about the single local WAN accelerator acting both as source and target for the same job. While the same server can serve as source WAN accelerator for one job and target WAN accelerator for another, it cannot be both source and target for one job. So you should add another local server in your set up for the cache to be created.

[mattmoore]

Hello,

We're trying to populate the Global WAN Cache on the Target WAN Accelerator, by first seeding it to a USB drive from the source WAN Accelerator, bring it onsite to the target cloud repo WAN Accelerator, and copying the data over. We'd like to then assign the WAN Accelerator to the Copy Job before its first incremental copy starts.

Would this plan of action work to accomplish this?

1) Change Source WAN to 20 GB to get a trg folder with only 20 GB of data.
2) Clear the Source WAN cache and re-populate it.
3) Deliver a USB and create a seed repo for it.
4) Create a copy job pointed to the seed repo. Disable it.
5) After the backup job runs, populate the source WAN cache.
6) Once the cache is populated, copy the trg folder to USB.
7) Enable the copy job and let it run to USB.
Bring the USB back and import the copy job and copy the trg folder to the target WAN Cache.
9) After import finished, rescan the cloud repo from the customer's VBR GUI, change the repo for copy job to cloud repo and enable WAN acceleration on it.

Thank you for your input.

Matt

[foggy]

Matt, the global cache seeding process is described in two first posts of this thread: you need to create a temporary local backup copy job with WAN acceleration enabled and then copy the cache folder to remote WAN accelerator. Source cache is not involved in this process.

[veremin]

By the way, have you already checked WAN Accelerator sizing recommendations? I'm wondering because 20 GB sounds a bit undersized. Thanks.

[pkelly_sts]

I'd like to do the above, moving from a pair of local disks to a higher-performing array of FC-disks, on the same physical server.

Is it as simple as physically moving the folder & repointing the config to the new location or is that going to cause any kind of cache refresh type processes to kick in (kind of like reading digests again I mean)?

[foggy]

Almost that simple, provided the absolute path to the folder is preserved. Don't forget to restart the service.

[pkelly_sts]

Ah, absolute path - you mean it has to remain on the same drive letter (in which case nothing has changed anyway obviously), or just the same path (in this case, the root) within the new drive?

Thanks for the service restart tip BTW, wouldn't have considered that.

[foggy]

Ah, absolute path - you mean it has to remain on the same drive letter (in which case nothing has changed anyway obviously), or just the same path (in this case, the root) within the new drive?

It should be the same path within the drive you've specified for cache while creating WAN accelerator.

Thanks for the service restart tip BTW, wouldn't have considered that.

Just to make sure, Veeam WAN Accelerator Service (not the backup service).

[lennis40]

We have run into problems here and there with the initial run of copy jobs using WAN accelerator. It's great once we get through the initial fingerprinting, but this seeding may be something we need to look into. If I understand this correctly, in order to seed the first full transfer using WAN accel, we need to have two WAN accelerators deployed at the customer location. Once the copy jobs completes using source and target, we copy the GlobalCache/trg folder from the target WAN accelerator.

We will likely copy this folder to the same USB drive we are using to seed the copy job as well. We will then copy the GlobalCache/trg folder to the same location on our target WAN accelerator, and then restart the services. Final step would be to change the target WAN accelerator on the tenant's copy job, correct? Just want to make sure my understanding of the process is correct, so we can speed these WAN accel jobs up for the initial run. Thanks.

[foggy]

Michael, your understanding of the seeding process is correct.

[sauterson]

I have read through various topics on this forum but i cannot seem to find a clear understanding of how to seed the WAN acceleration.
I am attempting to seed a backup copy job in the quickest way possible.

I am currently running a backup copy job using WAN acceleration. I have chosen the source and target wan accelerators. I am using a local repository for the first backup file thinking this would be faster. My plan was to then copy the backup file to a usb drive and ship it to the target location. That is not the case. It has been running for 47 hours and is only 11% done.

My question is. Should i run the backup copy job the first time without WAN acceleration turned on? And still use a local repository to create the initial backup file. Then copy it to a usb drive and ship it to the target location? Then get the backup file copied over to the target repository and remap the backup copy job. Edit the backup copy job and turn on WAN acceleration for the next time it runs?

Is it possible to select the source WAN accelerator as the source and target? Then copy the globalcache folder to a usb drive and ship it to the target location? Or should i setup a separate server as a target WAN accelerator then copy the globalcache folder?

[foggy]

Hi Scott, seems you'd like to seed both the backup itself and the cache as well, right? This thread answers your both questions regarding cache seeding:

- to seed the cache, you need to copy the 'GlobalCache\trg' folder to the target WAN accelerator (preserving the path) and restart the Veeam WAN Accelerator Service;
- single WAN accelerator cannot act both as source and target for the same job.

To seed the backup itself, you need to copy it to the target repository and map the job - your understanding here is correct.