Thanks for the interesting article.
"This is a true cluster". Yes you're right. And I think the cluster probably starts scaling even more when you have 4, 6, 8 , or 16 nodes. Something you can't do on MS SQL.
SQL Server Cluster is just a fail over solution, (or hot stand by), where SQL resources are switched to another physical box. But SQL resources are not shared among the 2 physical boxes.
Mysql also offers a fail over solution (DRDB i think). We're using Veritas cluster on Unix.
We also have a 64 bit, 16 CPU, 32 GB RAM and a SAN. CPU is sleeping all day long, yawning.
Yes, it's fast, for FullText search. Not meant for PK look up. (innoDB is faster for this).
I would also have written something about replication.
MS SQL replication is just pure pain. Mysql is so easier to set up and offer lots of features like
1 master, 1,2, 10 slaves, 1 master 1 slave which replicates to other slaves. Another nice things is you can have innoDB on master (for the writes) then MyIsam on the slave (for the reads).
The Web is 80-90% read in general. We're also converting our TextML server to myIsam until we find better solution on Unix.
I would have loved to see your hardware configuration.
Did you use any 64 bit?
Anyway, great article. I think Mysql is really getting some traction in big corpo.
Sr SQL DB (10+ years) and MySQL DBA (3 years)