Oracle Exalogic [EOL] Reviews

My experience with Exalogic was as an implementation engineer, and infrastructure architect for a deployment using two Exalogic X4-2 deployments alongside Exadata X2-8's. Across both datacenters, we had nothing but problems - random failures, random reboots, poor performance, and not one advertised metric or feature performing as promised. Notable issues were that the computer nodes would crash for a random, or no, reason at all, taking down the entire node and every VM on it, despite HA/live migration settings being configured. The entire set up was done by Oracle certified consultants, and Oracle leadership was involved in troubleshooting the issues.

I can't think of a single significant issue with performance or stability that was solved in less than nine months, and it is honestly the worst product I have ever dealt with. If anything, a "turnkey" solution should at least WORK when the company selling it are the ones who set it up, but my team and I consistently had to teach people that didn't even know the difference between KornShell and BASH BASIC concepts. The VM's only run Oracle Enterprise Linux 5.6, so there's massive security holes out of the box, and the virtualization performance of these VM's is atrocious. The disk read/write caps out at about 40 Mb/s on the VM's because of the eight year old NFS drivers they use to power the back end. They leverage Infiniband technology, but no kernel tuning has been performed at all. We managed to get acceptable performance after hours of my own team's troubleshooting and development, and not a single thing that Oracle did during the entire process was satisfactory.

Hardware failures for this major client were subject to three weeks or more of computer nodes being out of service because Oracle's service techs didn't stock parts in advance. Predictive failure never worked once. The replication of the ZFS system was at a performance level of something five years older, with numerous issues and bugs. I lost track of the number of times something "Shouldn't have happened" when Oracle came to us with excuses.

This happened in two distinct data centers with mirrored setups, so I know it isn't a fluke. This is NOT a product someone wants to deploy in a business that requires reliability and accountability. Oracle didn't deliver on a single advertised feature working properly, and if not for my team, our projects wouldn't have even launched. The admin teams had to work 80+ hours a week just to keep the hardware from self-destructing due to the incredibly shoddy programming, support, and configuration. That's not a "turnkey" system, that's garbage.