While over MyNoSQL we love all things NoSQL, I still think that scaling RDBMSes is an interesting topic some of the techniques used being the same in the NoSQL space. PCPro has published ☞ an article signed by Ian Wrigley that talks about the various steps for scaling a relational database (each step is accompanied by a short description and a nice diagram):
- master - slave
- master - master, plus master - slave
- horizontal partitioning (sharding)
- vertical partitioning
- transition to NoSQL (nb: this is actually my note. Anyway, please do keep in mind that scalability should not be the only reason for taking a look at NoSQL solutions)
In my opinion, the part about NoSQL itself is not as good or consistent as the first part on scaling RDBMSes and so I have found the title a bit misleading.