Your steps echo as you walk through the vast cavernous space. It is preternaturally quiet and cool. What was this space? Your companion whispers it used to be full of computers and the constant racket from fans trying to keep things from overheating. Is this the end result of serverless computing?
You awake with a start. You must have dozed off while catching up on the latest articles on cloud computing. Nervously, you look about to see if anyone has caught you napping in your cubicle. Looks all clear. You chide yourself on having that big pastrami sandwich for lunch. Thank goodness, you did not have to be at meeting – you would have been a face plant at the conference table. Now, what were you reading? Oh yes, serverless computing. That sounds like an oxymoron. Isn’t a server a computer?
Welcome to the growing magic of the cloud. First, it relieved you of physically buying and setting up the elements of compute, storage and networking needed to run your applications. You just sat down at your screen, logged in and ordered up the capacity and performance you thought you would need. You paid for it by the hour or by capacity used and could change or shut it down when needed. Plus, with a few more commands, it automatically scaled to meet demand, failed over if the hardware crapped out and all kinds of other things that were always a pain to set up and manage in your own data center.
Development and deployment times shrank, speed to market improved and all was good. The only drawback was you were still in the commodity plumbing business. You still had to specify the underlying infrastructure and you had to optimize for cost and performance. This is kind of a pain when all the value to users is in creating and running code.
Now, you don’t need to worry about the infrastructure at all. Just write your program and using some new services, the cloud provisions the right kind of infrastructure needed to cost-effectively support your application. Amazon Web Services (AWS) kicked this off in November of 2014 with its introduction of Lambda. Since then it has been growing in increasing popularity and adoption. Recently, the other two of the big three cloud providers responded with their serverless offerings. Google has Google Cloud Functions and Microsoft has Azure Functions.
It’s the next stage in cloud evolution where the machines are “smart” enough to know what is needed and they execute the provisioning.
And, take a look at how it’s priced. It’s an extension of the utility model but taken to its ultimate. You don’t rent a server by the hour, hence the term “serverless computing” or “serverless architecture”. Oh, they are still there, but you don’t need or care to see them. You pay only for the actual compute time you consume. It’s priced in billionths of a penny for every 100 milliseconds you use. Some claim seeing savings of an order of magnitude (i.e. 10X) over current cloud usage.
Computing is becoming even less expensive, faster and easier to use. Coincidentally, just about the time AWS introduced Lambda, I wrote a blog post about how cloud computing seemed to be in a race to zero regarding its cost and ease of use. I think I may actually have been too conservative. The pace of innovation and what it enables as a general purpose technology is just withering! A fair question is: What could be next?
But let’s be pragmatic. Is the key question for you: What are you and your firm doing with cloud? Or perhaps better yet, look over your shoulder. What are your competitors doing?