Feature request: separate locations for WAN Accel cache

[dgapinski]

Hello,

I am hoping that a good solution might be put into place for WAN Acceleration where caching is concerned. I've noticed that if I specify a cache amount of 500GB on an SSD drive with 950GB free, then the WAN Accelerator cache fills the entire free space (I've since bumped the value to 200 and am seeing that it takes up a total of 400GB of drive space. By that effect, the current setting is somewhat black art/meaningless because it implies an upper limit of space that admins like me would want to (or simply can) allocate. So would it be possible to build in a feature in a later version that either adheres to the upper limit or gives an option for overflow caching data to go?

My ticket for this is 952453
Thanks for listening!
Dan

[Andreas Neufert]

WAN Acc Target Cache is build for each wan acc connection.
For example 4 branches that copy data to HQ need 400 GB Cache when you set target cache size to 100GB.
The cache is reserved by the wan acc if used or not.

Then you need additional 18GB per 1TB used VM Source data for metadata.
So in our example if all branches have 4 TB data you need

4x100GB + 4x18GB Cache

[Gostev]

That is not correct Andreas, you don't add these two numbers together. Using your example, 100GB is required on target WAN accelerator ONLY, while 18GB is required on source WAN accelerator ONLY. Our System Requirements are actually very explicit about this difference.

Dan, please work with our support to identify why your target WAN uses more disk space for cache than specified in the wizard, as this is not normal. Thanks!

[cody.ault]

Hey Anton, we see a lot of confusion in support about this. I might point out that the following page matches what Andreas stated http://helpcenter.veeam.com/backup/80/v ... _many.html
After discussing this internally with one of our other Solution Architects, my current understanding of how this works(and I might be wrong) is when you specify the WAN Cache on the target of a Many to One scenario is that each Source WAN Accelerator will get a folder equal to the size specified. In Dan's example above, he specified 200 and it took up 400, so he likely has 2 Source WAN Accelerators and that's why he ran out of space when he used 500 as 500 X 2 = 1TB~. It seems like the better way to calculate the Target WAN Accelerator Cache size in the WAN cache settings is to take the total space you want to allocate and divide by the number of source WAN Accelerators using it and leave some room for updates. For instance if DAN wants to use 900 GB of his 950 GB drive, he should set the target Cache size to 450 GB if he has 2 source WAN Accelerators using that target. If he had 3, he should use 300 GB as the Cache size. Does that sound right?

[Gostev]

Yes, this part is correct and not what I had issues with. Incorrect was summing requirements of source and target WAN accelerators (100GB + 18GB).

[Andreas Neufert]

Did he used "populate cache" function there?

[dgapinski]

Hi Andreas. I did populate the cache first.

[Andreas Neufert]

As you see double cache usage I think the following happened:

Populate Cache => Cache was created.
Usage of WAN Accelerator => Cache for the pair of WAN Accelerators was initialized and reserved.

I don´t know if the "Populate Cache folder" was copied at WAN Accelerator Cache for the pair of WAN Accelerator connection.
If so maybe it is an option to delete the Populate Cache folder.

Anton, can you please help here?