This looks like the kind of product a website might use to scale up LAMP while increasing reliability. The benchmark Oracle ran on 8 servers provided 1 billion queries per minute and could do 110 million updates per minute from RAM.
“In 2002 we passed the limit of 1M reads per second. Now we’ve passed the milestone of 1B reads per minute. We achieved 1.05BN reads per minute on an 8-node cluster using MySQL Cluster 7.2.5.”
The scalability has been greatly increased:
“We’re very proud of those scalability enhancements that have made it possible to scale CPU usage per data node to more than 5x of what is possible in MySQL Cluster 7.1. In addition we’ve removed a number of bottlenecks making it possible to scale per data node performance by even more than 5x. Comparing our benchmark numbers for MySQL Cluster 7.2 we can see that we achieved 2.1M reads per second on 7.1 and we’ve achieved 17.6M reads per second, both on 8 data node set-ups using the same HW. Thus more than 8X better performance per node for MySQL Cluster 7.2 compared to 7.1.”
Even those who just add a second server for reliability might find it useful. Anything that increases throughput is useful because it doesn’t take a very complex query to bog down a single server. Perhaps the acquisition by Oracle may actually make the world a better place… Too bad they didn’t treat Java and OpenOffice.org with the same loving care.