I'd like to consider the use of DRBD for data replication between a primary and secondary location. The initial plan is to establish a VPN tunnel between the two; the primary end using a slice of a dual T1 link and the secondary location setting on a cable / dsl line.
The secondary will exist for DR only – it will "never" replicate back to the primary.
Has anyone done/tired/using something like this and what are your experiences with it.
Linbit also has a (Paid) DRBD Proxy product that is supposed to be designed for operating across WAN type links (compression, multiple peers). Anyone tried this out?
Best Answer
I can't speak for DRBD Proxy, but the regular DRBD will not like this much.
With even limited activity, you could easily saturate a dual T1 (2x 1.5Mbps; for round numbers, 300KB/s). 300KB/s could be taken up by application logging alone, let alone doing anything interesting on your server. This rules out synchronous replication (Protocol C), let alone adding the over-the-vpn latency into the equation.
Async replication (Protocol A) might technically work, but I would expect the secondary to be so far out of date as to not be usable in the case of a failure (the replica might be hours behind during the day)
Memory synchronous (Protocol B) won't help as it's still constrained by the bandwidth issue.
I expect DRBD Proxy will still suffer with similar issues, primarily causing replication delay due to the limited bandwidth.
I recommend you re-evaluate your DR strategy to work out what you are mitigating against; hardware failure or site failure.
In the case of protecting against site failure, you may get better mileage out of lower bandwidth / higher density transfers in the case of one (or both) bandwidth constrained site. Some examples of this technique are rsync (over-the-wire transfers are limited to changes in files between runs - rather than per-change for every change - plus some protocol overhead; can be run over SSH to encrypt and compress traffic further) and database log shipping (transferring compressed database logs to replay on the DR box may use less bandwidth than transferring a full database dump).
If you're protecting against hardware failure, a local DRBD replica connected with a GigE crossover will work just fine, allow for fully synchronous updates, and permit online verification to prove the data is consistent on both nodes. You can still combine this option with limited file replication to an DR site to protect against a primary site failure.