WAN Accelerator Cache Size - Growing out of control!

[wa15]

Hello Veeam Community,

So we have a many-to-one WAN accelerator deployment; with 3 of our sites doing backup copy job of 1TB worth of backups back to headquarters. At each remote site, the amount of cache is set to 50GB and at the corporate site, the amount of cache specified is set to 300GB. However, I am seeing that the WAN Cache folder at the headquarters is hitting 800GB! Any idea why it would surpass the 300GB set limit?

Any help will be appreciated

[Vitaliy S.]

Hello Farshid,

The cache size is configured per each pair of proxy servers. Since you have 3 to 1 configuration with 300 GBs limit, eventually you will hit 900 GBs (3 x 300 GBs) of cache size at HQ.

Thank you!

[wa15]

Thank you Vitaliy.

Looks like I have misunderstood how the size limit works. My understanding was that if I had 50GB allocated to each of the three remote site WAN accelerators, and 300 GB allocated to the HQ WAN Accelerator, then the maximum amount of space used would be 3 X 50GB or 150GB. But looks like that is not the case, correct?

So in my case, if I want to keep things under 300GB, I would really set the HQ WAN Accelerator cache size to 100GB so that it wouldn't exceed 300GB (3 proxies X 100GB). Is that correct?

Thank you for your help (:

[tsightler]

You'll also need some additional space for metadata, I believe the recommendation is 20GB per 1TB of source data so you'll want to do that math and subtract that overhead from the 300GB before setting the target cache size.

[Gostev]

Correct. The disk space requirements for WAN Accelerator are documented in the System Requirements section of the Release Notes document.

[wa15]

Thank you all for your help (:

[EddyChong]

Hi,

Sorry for bringing up the old topic but I would like to know how do we size of the source wan accelerator?

1. Does it have to be same size as the target wan accelerator cache size? Example the default 100GB.
2. Bigger size on source wan accelerator gives better reduction?


I know how target wan accelerator global cache sized but i am wondering how do we size for source wan accelerator. My scenario will be many to one.

Thank you

[veremin]

There is no such thing as source global cache, so, settings specified on a WAN Accelerator that is used as a source one doesn't really matter.

[EddyChong]

There is no such thing as source global cache, so, settings specified on a WAN Accelerator that is used as a source one doesn't really matter.

Hi Vladimir Eremin,

I see. I understand that digest files at source site will use some HDD spaces we specified in the source wan accelerator. In my scenario each source size has about 4TB data replicating by Veeam. Is it safe to assume that the digest files does not take much space so 100GB for source wan accelerator will be good enough? Bigger size at source wan accelerator will not improve the wan acceleration performance either?

Then if my deployment has 3 sites, total 12TB, I've specified global cache size at target 200GB, with 20GB for every TB, the actual hdd space required for global cache will be (3*200) + (12*20) = 840GB ?

Thank you

[foggy]

Digests require 20 GB of free space per 1 TB of source VM data, however, this is on top of the cache size, as far as I remember. You calculations for the target WAN are correct.

[EddyChong]

Hi Alexender,

Thanks for the confirmation on the target sizing.

Digests require 20 GB of free space per 1 TB of source VM data, however, this is on top of the cache size, as far as I remember. You calculations for the target WAN are correct.

What you mentioned regarding digest is it also applied to source wan accelerator? I'm a bit confused since all sizing calculation focused on target cache.

From the console while we configure the source cache size, the description "10GB for each operating system used in the environment", my environment only has Windows 2012 R2 and data is 4TB, so my cache size for source will be 10+80= 90GB ?

Thank you

[foggy]

What you mentioned regarding digest is it also applied to source wan accelerator?

Correct.

From the console while we configure the source cache size, the description "10GB for each operating system used in the environment", my environment only has Windows 2012 R2 and data is 4TB, so my cache size for source will be 10+80= 90GB ?

Since global cache is not leveraged by the source WAN accelerator, there's no need to allocate space for it (you can just leave the default, this space will not be used). You'd just need 80GB for digests on the source WAN server.

[EddyChong]

Hi Alexander,

Thanks for the clarification! Very much appreciated.