WAN accelerator - Is it relevant & improving performance?

[RichardH]

Hi
I’ve got a few WAN accelerator related questions:
1. It’s a 10/100 link between our live and DR site. Should I be using the WAN Accelerator?
2. Are there any jobs which it shouldn’t be used with? For instance, one of my servers is a 1tb SQL database server, with the large majority of changes each day being a 300Gb SQL backup file. Would this benefit from WAN accelerator, and if not would I be better off disabling it on this job to free up the cache for other smaller servers?
3. The bottleneck is typically ‘Target WAN’ – does this refer to the Target WAN accelerator?
4. What kind of storage should be used for the WAN Accelerator? At the moment it only has a 50Gb cache due to disk size limitations on the backup proxy at the DR site, however I could easily connect a larger USB2 disk or would this actually impact the speed negatively?
5. At the moment, only one backup job can be running the WAN Accelerator at once. Can this be expanded?
I appreciate some of these questions are probably unique to our site, however in getting caught up of the excitement of this new feature I don’t think I’ve fully understood, or can find answers to, whether it is actually relevant.
Thanks!

[veremin]

1. It’s a 10/100 link between our live and DR site. Should I be using the WAN Accelerator?

It was specifically designed for limited bandwidth (100mb or less). So, your deployment looks like the exact use case of WAN Accelerator.

2. Are there any jobs which it shouldn’t be used with? For instance, one of my servers is a 1tb SQL database server, with the large majority of changes each day being a 300Gb SQL backup file. Would this benefit from WAN accelerator, and if not would I be better off disabling it on this job to free up the cache for other smaller servers?

Yep, it would. Since WAN Accelerator uses much smaller block compared to typical backup job, you’re likely to see significant reduction in the amount of data sent through the bandwidth. Wan acceleration built-in capability makes it possible to “extract” and transfer only actually changed bytes. More information regarding it, can be found in this thread.

3. The bottleneck is typically ‘Target WAN’ – does this refer to the Target WAN accelerator?

Yes, it does. Though, it ok, because in overwhelming majority of cases the target global cache is the primary bottleneck in the WAN accelerator process.

4. What kind of storage should be used for the WAN Accelerator? At the moment it only has a 50Gb cache due to disk size limitations on the backup proxy at the DR site, however I could easily connect a larger USB2 disk or would this actually impact the speed negatively?

Storage with enough capacity is an ideal candidate for the target of backup copy job. It’s device actual capacity that should matted in this case, not performance, because the device will be mostly used for long-time archival purposes. Former production SAN, deduplicating appliances, etc. are considered to be good choices.

5. At the moment, only one backup job can be running the WAN Accelerator at once. Can this be expanded?

In order to increase number of concurrent tasks you might to put into use additional backup copy jobs along with additional WAN Accelerators. Thanks.