High Availability: Today!
Yesterday i was watching the video of the presentadion High Availability: Today and Tomorrow, presented @ SF in the first OHAC summit (05/31/2009). It was a really good discussion, and was really nice to see many different perspectives about HA.
But i think is really interesting what people thinks about HA, and if they use some kind of HA or not. The point is that with all the softwares we have to implement HA, and talking specifically about Solaris/OpenSolaris, and SunCluster/OHAC i know really few people that has some kind of HA on their sites.
In the different companies i did work, a few had HA; and 99% that were using HA, were using for Oracle, RAC. In GNU/Linux and *BSD, i think is way more complicated… we have DRBD, LVS, good softwares, but not so many implementations in my point of view.
We are having a good discussion about the project colorado on the OHAC community, and i’m talking about this because i think the great problem about HA implementions is cost. HA is one of that “tabus” on IT that people still think that is expensive, complicated, and is not for them. I did see many companies with problems with emails (pop, smtp, etc), with service downtime… when a simple LVS solution would give them HA at (almost) no cost.
Now the project colorado is here, and there is a lot of work to bring many features to give the users a chance to implement HA at low cost, and with a great focus on hardware minimization. There is an effort to build a solution to implement a shared-nothing cluster, and everything integrated on OpenSolaris. So, what is your opinion?
– Do you have HA in your site (RAC does not count ;)?
– Do you think HA is expensive?
– Did you try OHAC already?
– Do you implement HA on GNU/Linux or *BSD? What do you think about HA softwares on these environments?
– Do you think HA is important, or other recovery plans are better?