My first career, from my university days until the mid-1990s, was as a behind-the-scenes meteorologist. I looked at thousands of satellite pictures of clouds, usually those produced by hurricanes. I also took two courses about clouds. Clouds are really interesting, mainly because water is really interesting. Did you know that a cumulus (low, puffy) cloud inland is radically different from a cumulus cloud over the ocean? It’s the salt.
Fourteen years later, a client asked me to create operational procedures and a staffing plan for a new product that they were going to run “in the cloud.” Who hasn’t heard of “cloud computing?” I knew that it was near its peak on Gartner’s Hype Cycle.
I knew about Software as a Service (SaaS), epitomized by SalesForce.com. And I knew that you could rent computing power and storage by the hour from companies like Amazon and Rackspace. The web page you’re reading was served up from a computer out there, somewhere. “Out there” is also called “the Cloud,” from the cloud symbol used on network diagrams to indicate the not-our-problem, it-just-(mostly)-works public Internet. What’s the big deal?
As it turns out, The Big Deal is not just that you can have a computer Out There Somewhere, it’s that you can have a few of them, or a lot of them, automatically, and pay for them by the hour, and write computer programs that create, modify, and destroy—computers! And storage, and switches and routers, and firewalls, and load balancers.
Software is the new Hardware.
As with everything, there’s good news and there’s bad news.
For Business People
The good news
For business people, it’s almost all good news. Your up-front capital costs just went way, way down. You don’t have to overbuy hardware or fear success. If business takes off, rent more computers, and have them in minutes, literally. There’s only a little bit of bad news, but it’s a people thing and not a technical thing.
For Technical People
The good news
For technical people, the good news is that you’re no longer hamstrung by hardware availability based on business constraints. Need a sandbox to try something new? You don’t have to worry about finding space on an existing server, and worry about whether or not your experiment will interfere with what’s already there. And when you’re done, you don’t have to clean it up. Just delete it. The whole computer! Need a test environment that looks like production? Rent one for just as long as you need it, and then give it back.
The bad news
The bad news is what they told Spider-Man. “With great power comes great opportunity to make a colossal mess,” or something like that. All the mistakes that you can make with software, you can now make with hardware, too! Computers all over the place, no one knows what they do, and no two alike, and being created ex nihilo by software you—yes, you!—wrote. It’s like the Sorcerer’s Apprentice from Fantasia! And you get a bill at the end of the month, every month, for the whole mess.
The way to prevent the bad news is to get disciplined without getting bureaucratic about it. Hey, if it were easy, everybody would do it.
An amazingly helpful guide for my return to IT infrastructure and the cloud has been a pair of books from the IT Process Institute:
The Visible Ops Handbook explains, in around 100 pages, how to get from a nerve-wracking, fire-fighting operation to a sane one—nice in its own right, and essential for cloud. You have to stop being a cowboy and get a lot of discipline:
- Change management—fast turnaround, but no exceptions,
- Configuration management, including finding all the one-offs,
- Standardization of configurations to eliminate one-offs, and
- Identify and eliminate root causes for every incident.
In return, you get your nights and weekends back.
Visible Ops Private Cloud explains how to keep your sanity when you move to a private cloud (hardware-as-software, only running on hardware you own). You need yet more discipline, and you’ll need to learn some new (but cool) tools:
- Design and configure services, not just individual servers,
- Dramatically increase monitoring, and
- Fully automate provisioning.
In return, you get amazing efficiency and reliability.
When ITPI comes out with a public cloud book, I’m buying it. Public cloud (Amazon owns the hardware) is even more difficult because you can’t actually see and monitor the hardware, at least not directly.
For Business People
The bad news
You’ll need to hire and pay technical people who are not just smart, but disciplined, and that means disciplined enough to stand up to you. In return, you’ll get amazing efficiency and nimbleness—what Gene Kim (of The Visible Ops Handbook) calls “Amazing Kung Fu.”