Wan Accelerator - See what is in the global cache

[stevenrodenburg1]

Hello,

I've been running copy-jobs through two Wan Accell. for a couple of days now and it's running fine. On the Windows server in the remote site, I installed the wan acc. + repository and dedicated a VMDK to each.
The global cache drive is 250GB in size and the global cache-size for this wan acc. in Veeam is set to 200GB. I noticed that as soon as the very first job started, all 200GB got "filled" immediately.
The source cache is set to 100GB and is only around half "filled" after these couple of days.

I noticed that in most jobs, what is obtained from the global cache, is tiny. A couple of megabytes maybe. Sometimes even nothing. As there are plenty of 2008 and 2012 servers, win10 and win7 desktops that get backed up, I expected a much higher "hit ratio". In the very beginning, it was in the gigabytes range.

I must says that the amount of data that gets tranferred has sunk massively since the very first run, which is expected and the goal of the wan accell in the first place.

It does not surprise me that the hit ratio has gone down so dramatically, because if the source side decides not to send something, why bother on the destination side. This is the whole purpose of the product after all.

My question is:
I would like to take a peek in the global cache as to see what's in it.
I wondered how the very first job ever, filled the global cache to it's defined limited of 200GB right away (and that amount of data was not transferred yet, not by a long shot) and why the source cache has been growing slowly and is still around 50% of it's 100GB capacity after close to a week of running daily jobs.

[PTide]

Hi,

I would like to take a peek in the global cache as to see what's in it.

It's not possible to see what's inside WAN cache as all data is held in a binary format.

I wondered how the very first job ever, filled the global cache to it's defined limited of 200GB right away (and that amount of data was not transferred yet, not by a long shot)

During the first run WAN Accelerator Service on the target WAN accelerator populates the global cache storage with data blocks from the copied file. Do you mean that the amount of data transferred during the very first run was smaller than the amount of data that has been put into the Global cache?

Thank you.

[stevenrodenburg1]

yes. The local and (remote)global caches were empty after the install (obviously). The drives in Windows said 100% free.
I then started the very first job and POOF ! the global cache, set to 200GB, was at 200 GB filled (200GB used on NTFS, the drive-size is 250GB). So almost instantaneous and within a few minutes of starting the very first job so of-course it had not transferred much yet (it's only a 5 MB/s link).

My guess is the global cache get "pre-allocated" with place-holder files or something, explaining the almost instantaneous jump from 0% to 100% filled.

My question is basically "how much of this 200GB global cache is REALLY filled with *actual* data". It would allow me to see if it's under- or over-sized etc. Kind of like seeing if there are many cache-misses, meaning the data had to fetched from the source site, meaning enlarging the global cache would help etc.
I understand the data in the caches is in binary form, but I was hoping there was a "real usage" indicator somewhere (as 200GB was surely not all "actual data" during the first minutes of the first job). And it still might not be. The local cache is more "honest" in the sense that the used capacity as seen in Windows is a good indicator of "actual usage". The local cache, set to 100GB, took several days to fill up and is still only at 82 GB.

[PTide]

The global cache space is pre-allocated for the sake of performance so in fact not all of your 200Gb are really filled with data. Source cache works differently than target cache so its size grows slowly as it gets filled with real data. Currently there is no way to know the real usage, however you can use Veeam ONE to get some useful statistics.

Thanks

[alanbolte]

The local cache, set to 100GB, took several days to fill up and is still only at 82 GB.

This is a common misunderstanding. There is no global cache stored on a source-side WAN accelerator. Most of the data stored on the source side is digests of the backup files on the target site, and of the target-side cache. You cannot control the size of this digest data, so it may be much smaller or much larger than the cache setting.

[stevenrodenburg1]

This is a common misunderstanding. There is no global cache stored on a source-side WAN accelerator.

Where did I say that the local cache is the global cache? That is right. I didn't. Ever.

[PTide]

UPDATE:

You can go to the host that holds the role of a target WAN accelerator and check Svc.VeeamWANSvc log file that is located in C:\Program Data\Veeam\Backup. In that log file search for a "Free size" string which represents the real consumption of the global cache.

Thanks

[stevenrodenburg1]

Thanks for the tip. I look at that Logfile:

On May 11th which is the oldest log, it says:
[11.05.2016 10:32:28] < 3272> wan| Free size: 6.8750 GB

today's log says:
[25.05.2016 03:32:14] < 1632> wan| Free size: 2.0000 GB

There was about 6.8 Gigs free in the cache on the 11th and now that's 2 GB. Meaning that of the 200GB global cache, 198 GB is consumed so the cache is utilized nicely and its not under-sized.