Mythbusting: SQL Server doesn’t scale

There are some stories and myths around SQL Server. Like a lot of myths, these stories may have some origins in truth, but that doesn’t mean they are true now. This series of blog posts will address some of the key myths and give you the facts around them.

The first myth on my list is that SQL Server doesn’t scale.

How to answer this depends on your definition of scale. This is a term that could be applied to different things. You could be talking about scale in terms of the size of the database or the number of processes or the number of concurrent users. Whichever your definition, there are case studies to answer it.

The Mediterranean Shipping Company has a data warehouse of shipping data with 22TB of data in it. That’s pretty well scaled if you’re talking size. It also handles 397000 transactions per minute. So scalability of transactions isn’t in question either.

A French web analytics company, AT Internet, monitors more than 350000 websites and 25billion pages per month, with around 20TB of data. Again, that sounds scaled to me.

But it’s not just about being capable of handling large volumes of data. The database has to be fast as well. SQL Server 2008 set the world record for 1TB data load in just under 30 minutes. The previous record holder was Oracle, with about 45 minutes. SQL also set the world record for price/performance (i.e. performing well but at a lower price).

If you want some really big numbers, as well as a name you might be familiar with, there’s Myspace. 130 million monthly users, with about 300000 new users a day, resulting in about 1Pb of data. And it all runs on SQL Server.

Myth: busted!