High Availability for Virtualized Environments
By RickRamsey-Oracle on Feb 18, 2010
Used to be availability was a big deal.
The product formerly known as Solaris Cluster (then called Sun cluster and now called Solaris Cluster again) was the answer to many a midnight prayer because it did such a great job of keeping your Solaris operating environment available even across geographical boundaries.
(BigAdmin's resource center for Sun Cluster administrators is in serious need of updating, but you can still find some useful links about Solaris Cluster there.)
Over the last few years virtualization eclipsed availability. Everyone's focus was on hardware consolidation. Solaris server? Linux server? Windows server? Who cares. We can run all the software we want on whichever server we like the most. And hardware salesmen can pound sand.
Well, guess what? If in the past you were running Need for Speed on one server and that server went down, you only lost one business critical application. But with Need for Speed running on Windows, Quake 4 running on Linux, and DOOM hosted on Solaris (with voodoo and black magic the way Sun engineers at Rocky Mountain Technology Center used to do it after hours), all part of a virtualized environment created on a Sun Oracle Database Machine without the IT manager's knowledge, you're talking serious unrest in the corner office if anything goes wrong.
So availability is a big deal again. In fact, an even bigger deal than it used to be.
We're working on an update to BigAdmin's virtualization resources for admins that will point you to resources that will help you figure out not only the best virtualization solution, but the best way to make that solution highly available.
In the meantime, you can take a look at how to solve one small part of the availability puzzle by reading this BigAdmin Feature Article by J. Randriam:
It describes how to configure a 3-node Solaris Cluster to run Oracle Real Application Clusters (RAC) / Cluster Ready Services (CRS) on a Sun Storage 7210 Unified Storage System. What you get with that is high availability for the database, the OS, and the storage system. Main sections include:
- How to configure an NFS filesystem from the Storage System GUI.
- How to mount the new filesystem from the cluster nodes.
- How to create an iSCSI LUN for each node.
- How to connect each node to its LUN.
- How to disable fencing.
- How to create a quorum device.
- How to configure the Oracle RAC/CRS device to run on the storage system while taking advantage of the high availability services provided by Sun Cluster.
BigAdmin has some basic publication support, now, so we'll start to publish more content in the next few days.
Thanks for hanging in there with us.