By Caleb Small, Senior Oracle DBA and RAC Consultant
Dan Morgan and I had the pleasure of evaluating an ODA and presenting our findings to the user groups in Victoria and Vancouver BC.†† My detailed notes and screen shots are available here:
and Danís notes are availabe here:
The ODA truely represents an excellent value in hardware, and unprecedented ease of deployment for software.† Never before have I been able to take a new, unfamiliar piece of hardware and build a working RAC cluster in two hours!
The machine comes with a "bare metal" installation of Oracle Linux.† From that point, you run a couple of scripts to set up the initial network configuration, then upload a single "install kit" available from Oracle Support.† Unpack the install kit, and run the ODA installation program.† About an hour later, you have a working RAC cluster with database.† It's that simple!
The install kit and subsequent ODA patches include Linux, Grid Infrastructure, Database, OEM, firmware and any other applicable patches in a single patch set.
For systems on isolated networks, the Automatic Service Request feature will not be available, but patching can be accomplished by simply downloading the patch, transferring it to the machine, and applying it.† Currently, ODA patches require a short term outage.† Rolling patches will be supported in the future.
The installation of third party and user developed scripts and agents (eg. NetBackup, data loaders, anti-virus, etc) IS supported.† This is a change from prior policy.† Standby Database deployment for DataGuard is as simple a running a script Ė as long as you have two ODAs of course!
On-going administration, monitoring, tuning, troubleshooting, etc of course still applies, but it is a pleasure working on an engineered system where all the pieces really do fit together properly.
Multiple databases are supported, meaning that a single ODA high availability deployment could support multiple applications.† Graduated CPU licensing means you only have to pay for the power you actually need.† With up to 12 cores per node and 4 terrabytes of triple redundant internal storage, this would be sufficient for most small to medium size deployments.†
Additional storage can be added with a varitey of plug and play ZFS storage appliances available from Oracle.† Numerous extra ports allow connection to a separate backup network for tape dirves, high speed redundant interfaces, etc.
The integrated lights-out management interface (ILOM) allows all operations - including initial deployment - to be carried out remotely.† The ILOM "Graceful Shutdown" option really does a graceful shutdown of the cluster and powers off the hardware.
We were truly impressed with the machine, as were almost one hundred others who attended the User Group demos in Vancouver and Victoria.† This unit is a game changer.† No longer will you by paying high-priced consultants like me to build RAC clusters.