Earlier in the month I read this post by David Cramer about why the cloud is not for you. He made a lot of valid points and some of them are the same reasons I ditched the "cloud" myself some time ago.
A few years ago when I started a web hosting company and an artificial support agent service I had rented cage space in a downtown LA datacenter. The DC was only a block away from our office. It was awesome! We had to buy the servers, switches, etc. and the monthly nut wasn't cheap, but we were in complete control. Up until that time I had always had server accounts on the networks for whatever company I was working for.
After I sold my piece of the two businesses I no longer had a cage, servers, etc. to play with. I was bummed. It sucked. I contacted Peer1 and rented out my own cage for the new company I had formed, Netlandish. Issue was, I now had to buy new hardware and cover the monthly cage and bandwidth fee and my new company wasn't making any money. Guess I jumped the gun a bit.
I came out of pocket and covered the expenses until I can get some money flowing into the new business. Luckily, that didn't take too long.
I quickly realized that driving into downtown to manage the hardware and normal maintenance was going to suck. It wasn't very often, but when I had to do so it was such a pain in the ass.
As my contract with Peer1 was coming to an end I decided not to renew and go into the cloud. What the hell right? It was all the rage and I wanted to run with the cool kids. The major downside was that none of the "cloud" solutions offered FreeBSD as an option. That was almost a deal killer for me but I moved forward anyways.
By this point I had 5 physical servers all churning along to run the new CartFreak product that my buddy Joe and I were running (and still run to this day), my personal site, a few client projects I had run and hosted, and a half dozen info product sites. I figured I'd save a nice chunk of change by moving to Amazon AWS.
Running the numbers I was a bit surprised because I wasn't really saving that much dough. I ended up saving roughly 10% compared to my Peer1 cage. Still, I didn't have to deal with the hardware anymore (speaking of which, I still have 5 servers sitting in my closet! hah) so that made it worth it.
At first I thought being on the cloud was cool. Not really sure why. I didn't do anything different. I didn't make use of the API or spin up instances on the fly. No auto scaling. Nothing cool at all.
I did, and still do, use AWS S3 service. It's inexpensive and rock solid in the grand scheme of things. Though there was a few large outages at one point. Which just reminded me of the time I was interviewed in ComputerWorld magazine for one of the outages. Read the article...
After about 6 months I decided to switch over to Rackspace. It would allow me to cut my AWS bill down by something like 30%. Saved me money and I wasn't stuck on any of the AWS specific features so why not.
Rackspace was cool but the service wasn't as solid as AWS and the support was iffy. Whenever I'd call about issues (mostly internal networking issues on their end) it was always a huge hassle to get it diagnosed and addressed. Though I have to admit, as time went on their support did improve. In fact, it got so good that I no longer had an issue with it. Still, I didn't feel like the solution was right...
Enter ARP Networks. They offer top notch VPS service for a great price and they even support FreeBSD! They claim their servers are never over sold and the support has been top notch. I ordered a test VPS and toyed around. The service was great so I slowly started migrating my entire setup there.
In the end I had better service, better support, a better operating system, and saved more money!
So for me, it boiled down to support and honestly, money. As cool as all the features that Rackspace and AWS offer, I personally had no need for any of it. Like David pointed out in his article I've never needed to spin up a new instance on the fly. With proper planning I can just request a new VPS and a few hours later, it's ready to rock.
So before you jump onto the cloud bandwagon make sure you think through your requirements, expected growth, capacity planning, etc. and be realistic with yourself. You'll save money and headaches in the long run.