Hacker News Reader

Comment by jedberg | Hacker News Reader

It has nothing to do with that.

It is simply that EC2 offers us the best bang for the buck for our requirements (one of the main ones being able to spin up a lot of iron very quickly).

View on HN

May I ask why you don't use another cheaper provider for fixed base load and reserve EC2 for peak load capacity? Is it that the overhead of running on two different providers not worth the money you'd save? Or is this something you've considered doing but haven't had the time to implement yet?

> Is it that the overhead of running on two different providers not worth the money you'd save? Or is this something you've considered doing but haven't had the time to implement yet?

Pretty much yes to both of these. With so few of us, the overhead of running it and moving to it just doesn't make sense at this time.

You don't seem to really trying. You could be moving everything to one host and still be saving money.

Surely the traffic/load isn't that variable? Seems like even a VPS provider (linode/slicehost/etc) would offer far more bang for the buck.

The pricing for bandwidth alone on EC2 is simply insane.

Actually, it isn't so much the traffic. It's when something like Cassandra suddenly stops working correctly and we need to take emergency action to keep the site up by, for example, doubling our memcached pool temporarily.

None of the VPS's can give us that many nodes that quickly.

Also, bandwidth is by far our lowest cost. It is less than 5% of our total bill.

>> "Also, bandwidth is by far our lowest cost. It is less than 5% of our total bill."

I can't see why Reddit should be a particularly high CPU usage website. So perhaps time to optimize and swap out some things...

Reply on news.ycombinator.com