One of the biggest buzzwords in industry right now is, of course, “The Cloud”.
But for all the positive marketing hype, I see just as much negative pushback, especially from the older IT crowd.
For example, I see this meme (or some variation thereof) a lot:
There’s no cloud, it’s just someone else’s datacenter.
My initial response on hearing this is, so what?
- It’s someone else’s sewer treatment center. That doesn’t mean you need to manage your own wastewater.
- It’s someone else’s router (somewhere). Someone else is much better at managing all that hardware between you and your favorite websites.
- It’s someone else’s meat processing facility. Economy of scale at it’s finest. And if you want more boutique, you can probably find a local butcher who will do a much better job of turning a cow into your game day burger patties that you ever will.
- It’s someone else’s oil refinery. Hooray for electric cars and all, but many cars are still on gasoline.
- It’s someone else’s calculator and ledger. Maybe you do your own taxes, but many small businesses outsource their accounting. Is this a bad thing? Someone else is probably better at doing taxes than you are, and certainly has the time while you’re trying to build your business.
I could go on all day.
My point is this; why do you allow some of your services to be outsourced? The reasons are both cost and quality.
Someone else can leverage economy of scale to bring you power at fractional costs. You don’t buy an entire airplane even if you fly 12 times a year. It’s still much less expensive to basically pay someone else to do all the hard stuff, like maintenance and piloting.
There was one company I worked at where we ran MapReduce data reductions quite frequently. It was an easily threadable problem set, so we spent many thousands of dollars to try to get as many processors available as possible, to thread. But even for the $10-12,000 range we only got 24 (a 4-cpu motherboard with hexacores). In someone’s cloud, I can basically get hundreds of CPUs for literally $1.50 an hour. So that job that took 8 hours with 24 cores, I can scale up and get it done in 30 minutes for just a couple of bucks. It will take a LOT of jobs before it would have been worth building that compute cluster.
And this doesn’t even count paying for power, datacenter, and even staff.
OK, you’re probably not going to want to hear this, but the truth? These large cloud service providers are probably better at running a datacenter than you are. They are literally world-class in the nitty-gritty, such as power optimization, layout design, network topology, and so forth. I’m not talking about just slapping together a rack and keeping some servers patched. I’m talking about ensuring that every aspect of the building, the room, the electrical wiring, the cooling, all of it is optimized for that computing environment. Because to be honest, you might be using a colo but if you’re a business with your hardware in-house, then it’s probably a room that’s been re-purposed so your air handlers and PDSs have been installed later on.
Note: I’m not talking about you if you’re a datacenter company; I’m sure you’re very good at building out your racks and pods. But if your business goal is a software-based product or service facing your customer, then put your talent and energy focus on the software, not on the infrastructure.
This doesn’t mean you don’t need good IT talent anymore, in fact quite the contrary. But instead of a network team, a firewall team, a server team, an Active Directory team, etc., you can have a single Cloud Team that manages your entire infrastructure. And that’s a lot less training and recruiting that you have to do, which allows you to focus on hiring a small amount of really good people and training them, instead of struggling to simply fill a bunch of empty seats.
Bottom line, just because it’s someone else’s datacenter doesn’t mean it’s a bad thing. If you choose correctly, you can both save money and also focus your time and energy on your product.