I wrote this to Brian Hitney after we’d briefly touched on cloud computing in my podcast with him last week. I thought I’d post it here for any reaction from the rest of the internet…
…
Another point on the economics that isn’t brought up is the cost of data transfer and data storage in the cloud, which is often more than you’d pay for ‘normal’ equivalents.
I’ve got a server that I lease and pay $84/month for. Included in that I get 750 gig of transfer.
With azure, 12 cents per compute hour for 24 hours x 30 days = $86 - about the same base rate. But… let’s say I’m pushing out between 150 and 200 gig of bandwidth per month. 200 gig at .15 cents extra is $30. Factoring in I store about 40 gig of data on that server, that’s another $6. To replicate my setup on Azure (or Amazon) would cost me about $38/month more, which is a 40% premium over what I can get in the ‘non-cloud’ marketplace. As my data needs scale up, the differential gets bigger. EC2 pricing is roughly equivalent for storage and xfer, but the base computer rate is cheaper for linux images than for windows images.
There’s a big play to push ‘cloud’, and given the markup, I can see why. The hardware is commodity, and is essentially a one-off investment, but the data xfer is the lifeblood, and I’m actually a bit fearful of putting so much reliance in the hands of just a few monolithic companies who will then charge us a premium to move our own data around.
If I had a huge amount of number crunching to do, without much data storage or xfer needs, current ‘cloud’ offerings make sense. And for potential adjunct service in a pinch, or for testing. But as a long term strategy investment, the pricing needs to change and/or there needs to be more value in the mix for my taste.
