WAN accel not adhering to cache size limit

[Scol]

Hi Nikita,

Thanks for the answer. I went through the configuration of a new WAN accelerator and I figured out how it worked...
It was a bit confusing.

About my 2nd question, can you tell me if I need as much cache on my source WAN accelerator than on the target one ?

Regards,

Raphael

[Shestakov]

Sure Raphael,
Having equal amount of cahche on both sides isn`t obligatory.
Thanks!

[foggy]

In fact, source WAN does not use cache at all.

[darkmannz]

Hi, an installation I look after has an issue where it ignores the size limit of its cache. It is set to a maximum of 100GB but happily consumes another odd 80GB until the disk fills.

The Veeam Backup and Replication version is 8.0.0.2030

Thanks in advance,

[foggy]

Dany, please review WAN cache sizing considerations given above and check whether you've allocated enough space for it.

[darkmannz]

That explains it, room for improvement there for the Veeam interface (maybe its in v9 ). So the WAN cache size is just an indication and not a limit. I'll look at adding storage to the server to act solely as the WAN cache.

[shadex12]

Greetings,

I am having so difficulty understanding what is going on with my WAN accelerator global cache size.

Here's my setup:

Cloud Connect Backup server
1 x WAN accelerator set for a size of 200 GB on a 300 GB disk. Currently, almost all the space is consumed with only 5.7 GB free. There is no other data stored on this disk.

Remote site 1:

Backing up two VMs totaling less than 1 TB. All the same OS. They have a global cache size of 50 GB. According to documentation to make this work I would need 50 GB + 20 GB = 70 GB of space on the target WAN accelerator cache (which is 190 GB).

Remote site 2:

Backing up 5 VMs totally about 60 GB of data. Two OS types. WAN cache is also set to 50 GB. 50 GB + 20 GB = 70 GB space.


So total space I should need on the target cache is 140 GB. I currently have 190 GB, which should be enough, yet the cache consumes well over that.

Now on remote site 2 when I run the copy job it tells me global cache is disabled due to insufficient space. What is going on here?

[Shestakov]

Hello Evan,
Additional 20 GB per 1 TB of source VM data is used for digests as stated in the User guide. Please review the thread for additional information.
Thanks!

[shadex12]

Nikita,

I have read this whole thread and I still don't understand what you mean.

In the above post I said I had set a global cache of 50 GB at the source, and that I had less than 1 TB of data, which would require 20 GB. At the target I ensure I had the 70 GB of data to accommodate this.

So did I not account for the 20 GB of additional storage? Please clarify.

[Shestakov]

In the above post I said I had set a global cache of 50 GB at the source

As was mentioned, source WAN Accelerator doesn't use global cache, so it should not grow at all if you don`t use it as a target WAN as well.

Could you explain your dataflows a bit please? Where you have the source VMs running and the initial backup, where you copy etc?
Thanks!

[rnt-guy]

Hello,

I've been reviewing this thread as well as the guide online. I'm sorry but this is still way too confusing. We've been struggling with cache sizing for over two months even with support's help.

The problem I see comes from terms and processes

Say we have 8 OS'es = 80GB, plus 100GB minimum to start.
And we have 5TB of data = 100GB (20GB per 1TB).
Do we set the SOURCE cache to 280GB?
If we set it to 280GB how much extra 'free' space does the drive need for digests?

Here's where the terms get hairy. There is a cache folder on the source drive called "GlobalCache" despite people saying that term corresponds only to the target cache. My system has this directory taking up 300GB which is what the Cache size field in the properties of the Source WAN wizard.

The tough part is that when you present a WAN disk to the server, such as 500GB, what you assign to the cache leaves most of the rest to be for the digests. It's this folder that fills up. I see no recommendation on how much space the digests folder needs. Does it need only 50GB? 300GB?

The error does not indicate if it's the digest folder that's filling up or actually the value assigned for cache isnt enough.

[alanbolte]

RNT-guy: Clear the source cache. Never populate it.

[Shestakov]

Source cache is not used unless you transfer data in both directions.
As for the cache size, 280GB is a fair number for the target accelerator.

[nickpoore]

I'm confused on the correct cache size to setup for my company.

I have three remote offices, and one main office.
Each remote office has a HyperV server.

Office1 has:
3x 2012R2 server
1x 2008R2 server
650GB used, 2TB provisioned storage.
It has a local 32GB WAN Accelerator (on SSD)

Office2 has:
2x 2012R2 server
1x Win7 workstation - on the server
2x Win10 workstation - backed up with endpoint.
200GB used, 400GB provisioned storage.
It has a local 32GB WAN Accelerator (on HDD)

Office3 has:
3x 2012 server
300GB used, 750GB provisioned storage.
It has a local 100GB WAN Accelerator (on HDD)

My CC server has 200GB cache on a 250GB SSD.

I keep getting warnings : Global cache is disabled due to insufficient cache size on target WAN Accelerator.

I appreciate that some of my source cache's might be small, but I've put HUGE amounts of resources for the CC WAN Accelerator, and it still reports insufficient size.

What kind of cache sizes should I be using??

Nick.

[foggy]

Nicholas, you can find WAN cache sizing considerations in this thread.

[alanbolte]

Nicholoas, it sounds like you might be trying to create 3x 200GB cache files on a 250GB disk, because you have 3 sources. Try setting the cache to 20 GB instead of 200.