There aren’t many phrases that make my blood curdle. “Build it and they will come” is one (they usually don’t) and my ultimate rallying cry for a startup/business/consultant to get my goat early in the morning, “We’ll put it in the cloud“.
This phrase fills me with dread, ill thought out and swooned like a true abdication of responsibility. In a more honest term it’s saying, “Oh we’ll stuff it up there and all our troubles will go away and the depreciation costs will vanish from the balance sheet too.”
Also, when I’m talking to startups the automatic assumption is that sticking “in the cloud” means sticking it on Amazon EC2. A costly exercise it can amount to, I’ve seen grown companies cry under the weight of integration hassle and costly consultants when keeping all their stuff on “normal” server hardware within a hosting centre would have done just fine.
Cloud computing seems to have gone from the technology that could spawn servers up and down based on computing demand to a service where you just host websites.
The poster child for elastic cloud computing in 2007 wasn’t Heroku, it was the New York Times. All the articles from 1851-1980 converted in to PDF documents for download. The numbers were pretty much unheard of in terms of processing. Forays in to Hadoop and MapReduce that was unchartered territory. It was exciting and fun. Since then Amazon have Elastic Map Reduce (EMR) for such things but it was EC2 instances created on demand. This was true cloud computing.
In the end the NYT tests spawned 100 server instances and process many tetrabytes of data over the 24 hour period. The cost savings were amazing for a job that only needed to be done once (though they had enough time to do it twice).
Now we host websites and pay over the odds for what are seemingly mediocre machines.