Well I finally have run into a situation that I couldn't explain very well until some research was performed about my EqualLogic SAN. Currently we have two members at our main site with an array offsite setup as a replication partner. Had an issue recently where we lost connectivity to our switches due to some power problems, the SAN went down, VM's..ugh. ..it was a messy 30 minutes or so.
Anywho..I was asked..well why didn't our SAN at our DR site pickup and start taking on the load..why didn't the vm's out at the DR site just start working..It's complicated.
For one, we don't have that piece of VMware setup and configured. BUT...why isn't my replication partner just taking on the data as it changes at the primary site. "We don't want a backup..we want failover", I was told. Replication doesn't work that way with Equallogic. It has to be configured and replication times have to be scheduled. I explained to my boss because of the way we have replication configured...if we must revert to the data at our DR site..it's only going to be recent to my last replica that was performed AND there is some manual intervention. You have to set the volumes online. Ok..so I had my volume collections scheduled to replicate every four hours. I assumed what needed to change and I called Dell/EqualLogic to confirm. Basically, to get a near copy I would have to adjust my replication schedule..the closest it lets you configure is every five minutes. The good thing is I'm not replicating the entire volume..only the changes so it's minimal. And..rather than me pound my circuit all at once for a few hours a day..I'm gently tapping it all day.
Thursday, April 16, 2009
Subscribe to:
Post Comments (Atom)
5 comments:
This is a pretty common way for storage systems to work: iSCSI or otherwise. Some systems have "continuous replication" that initiates as soon as there is a change on the volume. But typically failover across a wide area network is hard to accomplish...especially since the link between the two sites is typically low bandwidth. One potential solution is setting up failover across a campus, and having 2 SANs connected via a high speed link. Or of course, keeping the data replicated as frequently as possible and making the remote volumes active....like you did.
chris,
how do you setup the equallogic box to replicate changes only? i've just inherited a couple of these boxes and don't know that much about them. any help would be greatly appreciated.
thanks!
Fowlerlfc -
Once you configure your replication partner and enable - replication takes care of the rest for you. The first replica is the complete volume. After that, only the changes/deltas replicate.
What you do need to configure is the local reserve and the replica reserve space. The local reserve is storage used to preserve the state of the volume while replication is in progress. The replica reserve is space on the box being replicated to that houses the replicated volumes.
When configuring replication you can choose how many replicas you want to keep. When you reach that limit, the old replicas will simply be removed.
Determining how much data on your volume changes is important when configuring replication as well as understanding bandwidth requirements to replicate your data.
Dell/EqualLogic has a great PDF on replication and sizing replication space that I'd encourage you to get.
We are in the process of migrating to VMware and Equallogic SANs and will be replicating the SANs. This is probably a stupid question, but with replication set up properly is there still a need for a backup solution such as VEEAM, et al?
Thanks!
@MrBreeze,
I use replication as a continuous process and only plan to use those replicas if I ever need to failover to our DR site or need to failback a volume (worst case). I'm only holding on to several replicas at the remote site and if replication is my only backup mechanism and I needed to restore further than my number of available replicas..this would make for some unhappy people.
So..my thoughts are incorporate your backups into your replication and snapshotting schedules. That is the approach I take.
--Replication offsite
--Hourly/Daily snapshots
--Backups to meet your recovery objectives.
Post a Comment