Everything you think you know about AWS is wrong

December 10th, 2016

If you are like most people, you probably thought something like this the first time you encountered Amazon Web Services (AWS):

“Wow. A completely virtualized data center in the Cloud? How cool is that?”

And then, you may have crunched the numbers, compared them to the cost of your on-premise servers or your legacy (physical) data center downtown, and thought:

“Whoa! I’m going to be paying how much to host my virtualized Windows servers at AWS? I can buy a physical server a lot cheaper for a one-time cost, and the data center will lease me a whole rack for a fixed monthly price, where I can host all the servers I will ever need. No thank you!”

I have known a number of intelligent, well-informed, and highly technical people who have thought the same thing. If this was you, then you would be in good company. You would also be wrong.

Please don’t be offended. I thought the same thing. If you try to do a straight apples-to-apples comparison between your legacy servers and AWS, then AWS does appear to be more expensive.

The reason I was wrong (and you are too) is that you can’t do a straight apples-to-apples comparison. AWS is not your legacy server farm, or your legacy data center. It doesn’t function like your physical servers, and you can’t price it the same way. Well you can, but if you do, you will miss the point — and probably cost yourself a lot of money in the long run.

Let me try to provide a helpful analogy. Some of you reading this may be old enough to remember when “spreadsheets” weren’t computer programs — they were rows and columns of numbers literally “spread” across the facing pages of an accounting ledger. There were obvious advantages to taking these bulky, awkward, physical artifacts and virtualizing them as digital representations which could be stored on a portable disk. Anyone who has ever used Microsoft Excel, however, knows that a modern spreadsheet is much more than this. With the advent of computer spreadsheets, users were able to embed into the sheets mathematical formulae they were previously required to solve themselves. Eventually, users were able to cross-link different spreadsheets with one another. These advances automated much of the repetitive calculation that went into an accounting ledger. This made computer spreadsheets an exponential, not a linear, advance over paper. Today, we can use spreadsheets to perform tasks that were impossible — not just impractical — before computers. The spreadsheet didn’t just make accounting operations more efficient — it completely changed the accounting landscape.

This is the scale of the change AWS is bringing to the IT industry. AWS is not a datacenter — or rather, AWS is not just a data center. You can make the argument that EC2, the AWS service for hosting server instances, is a virtual data-center — but that argument still misses the point. EC2 instances are powerful, not because they are in the Cloud, but because (in the words of a friend of mine) these servers are code. They are ephemeral, to be created for a particular purpose and destroyed as needed when that purpose has been fulfilled. This is so far beyond the capabilities of a traditional data center it makes the move from accounting ledgers to computer spreadsheets appear trivial by comparison.

The flexibility and power of EC2, however, is just the tip of the AWS iceberg. EC2 is only one of dozens of AWS services, most of which have nothing to do with hosting server instances, and which cannot be compared in any way to a data center:

  • Relational Database Service: Host your MS SQL, MySQL or Oracle database directly, no server required (of course there is a server, but you never see it — which means you don’t have to install, configure, maintain, or manage it.)
  • RedShift: Data-warehousing service allowing you to store petabytes of information for the purpose of in-depth analytics
  • Quicksight: Carry out the in-depth analytics appropriate to your organization on the data you have stored.
  • Elastic Beanstalk: Build scalable Web applications.
  • CloudFront: Efficiently distribute your scalable Web applications to a global user-base.
  • CloudWatch: Monitor your usage across the wide array of AWS services.

This is just a small sampling of the AWS ecosystem. These services can tie-into one another, leveraging their various capabilities to produce immensely powerful operational environments. Some services eliminate the need for entire classes of server — reducing (or potentially even eliminating) the need for your datacenter, rather than just moving it into the Cloud.

It’s provocative to say “Everything you know about AWS is wrong.” This is obviously an exaggeration. Perhaps, however, the way you are thinking about AWS is wrong. The words “paradigm shift” get thrown around too readily, but they are the simple truth in this case. AWS is positioned to eliminate datacenters, and even servers, at the level of the individual organization. It provides companies of all sizes with capabilities that were formerly available to only the largest enterprises — or in some cases weren’t available at all, because those capabilities didn’t exist.

The sooner we recognize the IT landscape has completely changed, the sooner we can take advantage of what the new IT world has to offer.

Or we could just keep doing our calculations on paper…

Brian S. Pauls has over 20 years of experience navigating the rapid pace of change in computers, networks, and the Internet. He has found that getting it wrong is often the quickest path to getting it right.

This is a companion piece to his article “Amazon is changing the world with a platform you’ve never heard of.”