In my previous blog post, I emphasized that internet scale design can be implemented for any type of company. Whether it’s a small, bootstrapped startup or a rapidly growing, well-funded tier 2. But if it’s suitable for that many companies, why isn’t everyone moving into the cloud? In my opinion, there are two reasons.
First, the model of utility computing doesn’t work for all business models. It is most effective in models where demand changes, where there are peaks and valleys for larger scale systems. It also works well as a way to get your startup or project off the ground with little-to-no capital investment. In the story I began in my previous blog post, the seasonality of their environment made them a perfect candidate.
The second is more of a people problem. In many companies, IT leadership, SysAdmins, Developers, DBAs, and everyone else involved in service management, have been working with whatever technology stack that company has been using for years. It’s important to remember that most SysAdmins see their primary job as keeping things up and running, so we typically prefer working with things we know vs things we don’t.
If a C-level executive or VP returns from a conference about cloud, and issues a mandate that they need to “move everything to the cloud!” to remain “competitive” the SysAdmins will likely fail. Why? Not because they’re not smart enough, but because they simply don’t know enough about it.
While it would be ideal for the COO to say, “I want us to look into moving our platform into AWS, so I’m going to send you to get Amazon certified,” it rarely happens. Usually it sounds more like, “You’re smart, you’ll figure it out. Oh sure, you already have a full-time job keeping the lights on, but just squeeze it in when you can. We’ll need to see a POC by the end of the quarter.”
I don’t need to tell you how this ends ? it’s obvious. It will fail almost every time.
One of the amazing benefits to the Pythian model is that our teams are exposed to a wide variety of systems. We have built auto-scaling systems in AWS, OpenStack systems, VMWare systems, as well as legacy physical systems we support. Our teams are not relegated to whichever tech stack they happened to be stuck with for the last five years.
The bottom line here is that it doesn’t matter what kind of company you’re at – Whether it’s a small retailer, midsize tier 3, or larger tier 2, if you’re willing to sign on for the concept of site reliability engineering and commit to it, together we can accomplish some amazing things, all for a price you can afford.