Wan Accelerator Cache Hit Rate?

[wa15]

Quick question all,

Is there a way to view how effective the Veeam wan accelerator cache is and what the cache hit rate looks like?

Thanks!

[Gostev]

Yes, as far as I remember this information is provided in the Backup Copy job session log. Keep in mind that it literally reflects global cache hits only (which is exactly what you want to see), and does not reflect savings from 2nd level cache (data from existing backup files in the target backup repository). Thanks!

[wa15]

Got it, thanks Gostev!

I am seeing a seemingly low global cache hit. Looking at a couple of 1TB backup copy jobs, I am seeing 70.8MB obtained from cache and 2.2GB over the network. Or 823MB over the network and 46MB from WAN accelerator cache. These are on a 50GB source cache and 150GB target/global cache. Are these "normal"?

Also, somewhat of a side question, if a regular backup job is made onto directly attached storage AND ALSO eventually being added to a backup copy job and "shipped" over the WAN, should the storage optimization under the job's advanced options be "WAN Target" or "Local Target"?

[tsightler]

A lot of it is based on the type of data and whether that data is likely to have commonality. For example, a server that's storing images or that is using Windows 2012 dedupe or NTFS compressing is unlikely to see significant reductions from the global cache. On the other hand, transactional servers that have smaller, more consistent changes can see fairly high hit rates.

The global cache and the secondary cache are designed to work together to provide the best level of reduction so I'm curious to hear what your overall data reduction is. For example, how does the amount of "Processed Data" compare to the amount of "Transferred Data" on the copy job. The "Processed Data" is basically equal to the total amount of changed data with no compression or other data reduction, while the "Transferred Data" is what it actually had to send over the wire and will include a reduction ratio beside it, that's really the number that indicates what WAN acceleration is doing for you.

[wa15]

Got it, thank you sir.

Just to answer my other question, if a regular backup job is made onto directly attached storage AND ALSO eventually being added to a backup copy job and "shipped" over the WAN, should the storage optimization under the job's advanced options be "WAN Target" or "Local Target"?

[Gostev]

This settings has no impact on WAN acceleration whatsoever, as the backup file is completely "disassembled" for processing anyway, and WAN acceleration then uses variable block size fingerprinting to find common data patterns.

[wa15]

Thank you Gostev.

Reason I asked is because I thought I saw somewhere that if a backup job is bound to be transferred over the WAN via a backup copy, you should use a "WAN Target" optimization so a smaller block size is picked. But looks like I was wrong. Thanks!

[Gostev]

What you definitely should have, is the same block size across all your primary backups. As soon as you have different "primary" backup jobs producing backup files with different block size, Backup Copy jobs that include VMs from different primary backup files will refuse to work.