If You Aren’t Using SSDs For Your Solutions, You’re Wasting Time and Money

I’ve been advocating the enterprise use of SSD/flash for a long time. They solve the #1 problem of big systems: storing and retrieving lots of small parcels of data.

So I thought this video with Artur Bergman, while a bit rough around the edges, is truth.

If you aren’t already using or at least seriously considering flash in your stack, you are wasting money, likely busy solving last-decade’s problems.

Ignore the FUDsters that make their presence known in every flash discussion.

No, SSDs aren’t unreliable. They’re surprisingly reliable outside of a couple of bleeding-edge desktop enthusiast brands. They aren’t particularly expensive.

And yes, IOPS is the strangle-point of almost any system that can’t be held in the confines of RAM. Every single “my database can’t join” argument is 100% caused by the crippling restrictions imposed by spinning rust.

SSDs change the entire storage discussion. The orders of magnitude jump in random-access performance demolish existing assumptions about data storage and retrieval.

They completely change the cost equation for normalization (heavy normalization can actually be a net performance benefit when using SSDs) and can signficantly influence caching patterns.

Even on the desktop, the prices of SSDs is just a no-brainer now. No, you don’t want to use them to store your movie collection, but for everything else it is just all winning.