AWS Cloud Computing Astonishment


Everyone uses Amazon.

Like me, a lot of people have Amazon Prime, which makes quick home delivery of just about any product you need a reality. But, even if you’ve managed to avoid that temptation, you’re still using Amazon; you just might not realize it. Amazon’s “AWS” cloud computing products launched in 2006, and you’d be hard pressed to use the Internet now without running into it somewhere. Amazon’s own case studies list Expedia, Pinterest, Netflix, Adobe, NASA, and that’s just scratching the surface.

Even if you are familiar with Amazon’s offerings in the space, what you might not be aware of is just how cost-effective cloud computing has become. Recently while conducting some performance testing on the application server that backs our mobile forms platform Mi-Forms, I had the opportunity to explore some of the larger and more exotic offerings. Going into these tests, I basically knew the pricing of a 2 core system with 8GB of memory. That’s the sort of machine we have set up both as our own cloud servers and for our customers. What I wasn’t prepared for was just how much computing resources were available.

Amazon’s EC2 has several tiers of computing resources available, and the system falls squarely in its general purpose computing, costing about 26 cents per hour (~$190/month without any commitment. But what if you want 32 cores and 60 GB of memory? This is the type of resource that in the not-so-distant past was only available to universities and huge corporations. Well, it turns out Amazon can spool up one of those for $3 per hour (~$2160/month), again without commitment. That may sound expensive, but think about what you could do with a machine that--by Amazon’s calculations--is equivalent to over 16 of the typical server machines I previously mentioned.

That type of machine is, of course, overkill for most applications. It  ran through our performance tests in times never seen before either with locally-hosted or cloud-based systems, and if we’re being honest, our test clients couldn’t keep up with the capacity it provided. From some creative searching and looking through references, it seems that this machine is on par from a raw processing power perspective of the Cray X-MP/48, which cost $15 million in 1984.

And of course pricing declines don’t only apply to CPU and memory resources. If you want to provision a quick terabyte of storage, that will only run you $50/month at the most expensive level. Want to throw in load balancing? No problem there either, as even a reasonably high volume site won’t cost more than about $20/month.

With what seems to be a “race to the bottom” in computing resource pricing, cloud computing options are becoming the norm rather than the exception. It’s a challenge to find a good reason to create a complex internal infrastructure rather than taking advantage of one provided for you. Amazon isn’t the only game in town either. Similar computing resources are available from Microsoft Azure for equal or lower cost. Mobile information collection and analysis solutions from Mi-Corporation benefit from resources available in the cloud from and may very well be the right solution to run your business more efficiently.

 Photo source: