13 reasons to ditch AWS for another cloud

Amazon beats the other public cloud providers in lots of ways, but not in every way

13 reasons to ditch AWS for another cloud
Thinkstock

Amazon Web Services clearly dominates the field. One of the first clouds, AWS is today’s leading choice for good reason. It offers so many options and services under its virtual roof that it’s nearly impossible to summarize the breadth. There are dozens of machine types to choose from, dozens of ways to store data, and hundreds if not thousands of software packages you can use to build out your environment. That’s the definition of the 800-pound gorilla in the cloud.

But the competition isn’t standing still. Not only are there great alternatives to AWS right now, but many of these are evolving in ways that hint at the future of the cloud—when AWS may no longer be your first choice.

So before you hit the default button and spin up servers on AWS, consider whether Amazon’s cloud is right for your project—now and in the years ahead. Here we take a look at 13 scenarios where going non-AWS makes sense now, and why you may soon dump AWS for another cloud in the near future.

You want bare metal performance

The cloud depends upon virtualization to deliver all of its flexibility, and virtualization comes at a price. Some little bit of code must look at every I/O request and send it to the right virtual machine. Those little bits add up. The folks at Joyent love to talk about how they cut out the middle step to offer “bare metal” performance for Docker containers on their machines. They spent the time developing Triton so they could remove a few layers that normally slow down I/O operations on virtual machines. Our tests showed that you could get more bang for the buck.

You want a dedicated machine

What we call “machines” in the cloud are for the most part just slices of bigger machines. We like to think we’re getting root on our own “box,” but it’s really just a shared server we’re accessing, pretending to be a separate machine. Do you trust your neighbors? Are you sure that the cloud programmers have stomped out all of the bugs? How paranoid do you feel? IBM Cloud offers independent, bare metal boxes that you configure with your own favorite specs. You choose the amount of RAM, SSD, and processor cores. Someone builds the machine and then it becomes yours and yours alone for a month, the standard length of a lease. If you need to rent by the hour, you can choose from some standard configurations available.

You’ve only got a few web pages

It’s been years since the early shared hosting arrangements appeared, letting people buy an account on a UNIX machine to maintain a few web pages served up by Apache. Sure, it only ran on one computer and it didn’t scale, but many, smaller websites didn’t need that flexibility. Many sites continue to run quite smoothly on WordPress, Drupal, and other simple options from an earlier, simpler time. Some people still write static web pages. These shared servers running CPanel are still perfectly good options that are often simpler and more efficient than the cloud—and often dramatically cheaper. If you’re just delivering some PHP or classic web pages, some of the older shared servers are the cheapest way to deliver your data.

You want solutions, not a platform

Office 365 continues to be a popular solution for everyone who needs to edit documents or create spreadsheets. G Suite has its own fan club as well. Amazon is a platform and thus an opportunity—not a solution. It’s entirely possible that the best way to solve your problem is to build an extension to the browser and store everything as documents in Google’s or Microsoft’s system. Or perhaps you have another way to leverage those online apps. Sometimes renting your own server and building out your own infrastructure is the last choice you should consider when fully fledged solutions already exist in the cloud, ready to serve you.

You want to write less code

App Engine was one of Google’s cleverest ideas. It lets you write some basic Python code (or Node.js, Java, Ruby, C#, PHP, or Go) using a very strict structure and then Google handles the scaling, the data storage, the configuration, the load balancing, and more. It’s quite easy to build a pretty complicated application with only a few lines of code that glue together Google’s resources. If you end up with a bazillion users, Google handles the scaling and bills you by how many of the resources you use.

You want a smart, distributed database

For years now, the cloud data storage world has been driven by the casual pack rats who are happy just to squirrel away some information without worrying about inconsistencies. They liked to talk about “eventual consistency” because, well, that was the best they could do. Now the pendulum is swinging back. Google Cloud Spanner is a highly scalable, replicated data store that offers strong consistency across all of the nodes. All of the nodes work together to keep their story straight. Strong consistency and global replication? Google is offering it.

You’re a data scientist not a developer

Amazon, Google, IBM, and Microsoft have some pretty smart experts in machine learning on their staff and they’re willing to sell you the tools on their clouds. Between them, they already have several dozen services that offer machine learning in many shapes and forms. All of them offer tools for cleaning your data and training and testing models, but Amazon takes a somewhat simpler approach. Those of us who skipped the higher math might appreciate the simplicity, but real data scientists may chafe at the more limited options. If you want a broader choice of models, or to tinker with the algorithms, one of the other clouds may be a better fit.

You want to use Google APIs

Google offers amazing APIs for services like translation and mapping. There are dozens of worthwhile Google APIs to add to your applications and they’re available to every server on the web. You don’t need to be on Google’s cloud to use them, but in many instances applications run faster and smoother when the machines are under the same roof. So if you’re working on projects that depend heavily on Google APIs and you don’t want to get bogged down, Google Cloud should be your go-to service of choice.

You want to use Windows

AWS offers machines running the Windows operating system, as do many other clouds. But there’s no place like home and that’s as true for Windows as anything else. Microsoft’s Azure cloud offers many flavors of Linux, but you can’t help but notice it has a soft spot for Windows. You can transfer your existing licenses when you move to Microsoft’s cloud and even juggle your licenses with a “hybrid” strategy, leaving some on premises and some in the cloud. If you’re running Microsoft virtual servers, Azure is ready for them.

You want to use Watson

IBM Watson’s Jeopardy win was impressive, ushering in a new era of interest in machine intelligence. If you want some of Watson’s limelight to shine on your project, you should turn to IBM. The company boasts among its ranks many computer scientists who have devoted their lives to creating artificial intelligence tools, and it is now marketing them under the Watson brand name. IBM’s Watson cloud offers solutions such as Virtual Agents, which help users by interacting with conversational English instead of annoying menus. But that’s only a start. A whole smorgasbord of analytics tools and models is available.

You like specialty data sources

One way cloud providers have been trying to differentiate themselves is by curating large collections of data so you don’t need to. Amazon has archived lots of weather data and this is just one of the data sets they’ve got ready for AWS users. The other cloud providers have their own specialty collections. IBM offers a searchable collection from Twitter’s PowerTrack stream with some extra features like the ability to “validate” a tweet by checking to see if it’s still published. Google has dozens of data sets ranging from sports (Major League Baseball) to cannabis genomics. Microsoft offers census data, airline on-time performance, and much, much more. Microsoft has even preloaded many of these troves into Azure Machine Learning Studio, its cloud-based machine learning toolkit. If one cloud has the data you need, it makes sense to put the machine running your analysis in the same cloud.

You want to analyze video

The IBM Watson brand name is turning into a big umbrella for almost anything IBM AI scientists create. One of the more intriguing offerings is video analysis, which leverages Watson APIs like speech to text, natural language understanding, and tone analyzer—and even live social media responses to the broadcast, if you like—to perform “live event analysis,” “video scene detection,” and “audience insights” on your video. So if you’re trying to figure out just why cat videos are so compelling, IBM’s cloud may be for you.

You want to play with blockchain

The world of crypto currencies is complex and full of experimentation, but it’s not all sketchy deals and anonymous transactions. IBM’s cloud is just one of the options for building a secure and trustworthy record of business transactions using the ideas from the world of Bitcoin. Your business records can be locked in virtual concrete and wrapped in metaphorical amber so everyone who comes along later can trust what was recorded long ago. These smart contracts can provide a stable foundation for future business by eliminating disputes.  

Copyright © 2017 IDG Communications, Inc.