Cloud Computing: A Necessity, Not an Option

Competition is a wonderful thing. And the increased competition in cloud services – especially the intense battle among Amazon, Google, and Microsoft – is resulting in services and pricing that would have been hard to imagine just a few years ago.

Over the past month, all three of the big players, plus a variety of smaller ones, have introduced new products and services for the cloud.

About a month ago, Google started this round of price cutting at its Cloud Platform Live event, where it cut prices on most of its services, some rather dramatically. Compute fees dropped by 32 percent, and standard storage prices went down to 2.6 cents per gigabyte per month. A similar drop in prices for its consumer product, Google Drive, gives you 1TB of storage for about $10 a month. That's quite compelling.

More importantly was the company's assertion that "We think cloud pricing should track Moore's Law," which suggests that pricing decreases will continue.

Not to be outdone, Amazon came back with its own price cuts on its services, including a 10 to 40 percent drop in the cost of most raw compute capabilities. Standard storage pricing starts at 3 cents per GB, decreasing with volume. Or, if you want less redundancy, the company's slower Glacier storage is just 1 cent per GB.

At its Build conference, Microsoft tried to show that its Azure platform was getting broader in terms of tools supports and ease of development, with new features such as a better way of managing a complete application in one place. It too cut standard pricing, with compute prices down by up to 35 percent, and storage by up to 65 percent. It also introduced a new less expensive "basic" level without features such as load-balancing and auto-scaling. In general, its pricing for compute is very competitive and it offers multiple options on storage depending on the amount of redundancy you want.

Actually figuring out which company has the best pricing is very much application-dependent, as you'll have to figure out how much storage, compute, data transfer, etc. you'll need. Of course, as one of the main draws of cloud computing is that you can scale it up and down as you need it, your price may vary each month depending on what you actually use. To that end, I was pleased to see Microsoft and Amazon recently improve their dashboards for tracking usage and cost.

And I'm also pretty impressed by each vendor's unique offerings. While all the vendors offer standard infrastructure services – compute, storage, SQL and NoSQL databases – and even a bit of "platform as a service," they all have somewhat different strengths.

In many ways, Amazon Web Services led the move to cloud services, and it has perhaps the broadest set of cloud offerings with things like Glacier storage, and its Cloudfront Content Delivery Network, Redshift Data warehouse, Kinesis stream processing, and Workspaces virtual desktop. I was recently intrigued by its newest R3 instances for applications that require a lot of memory – for instance, you can now get a version with 32 cores, 244GB of RAM, and two 320GB SSDs for $2.80 an hour. That's overkill for most applications, but I can easily see a company that has an occasional need for memory-intensive applications choosing this rather than buying an expensive server.

Google is stressing its cloud-scale applications, with a recent emphasis on things like its BigQuery data analysis tool aimed at multi-terabyte datasets. Google has a lot of tools for developers, and is stressing how well its Cloud Platform fits in with those tools, especially for connections to Android apps.

Microsoft's big difference with Azure may be the way it connects with on-premises solutions, and pushes solutions that connect cloud and on-premises solutions through things such as Active Directory federation. In addition, it is stressing its long connection with enterprise developers through things such as extending its Visual Studio and .NET tools to Azure. At Build earlier this month, the Windows and enterprise developers I talked to were quite impressed, though the ones who primarily make iOS and Android apps and thus use more open source tools were a lot more skeptical.

And of course, there are lots of other choices. In particular, the OpenStack movement led by Rackspace and others is getting a lot of attention for being a more portable option, particularly for larger enterprises that want to run their own private clouds and perhaps federate these with public cloud services. Most of the users I've talked to are still in the prototype stage, but lots of companies have high hopes for this platform, and I've been impressed by the amount of support the project seems to have.

For instance, HP recently announced its new Cloud System aimed primarily at enterprise customers, with new automation and management tools, and a focus on integrating public and private cloud services. This is built on HP's implementation of OpenStack. The company has been focused mainly on hybrid cloud implementations, believing that is how most enterprises will run in the future.

IBM as well is making a much larger push with its cloud services, notably its acquisition of Softlayer for cloud infrastructure and Cloudant for database as a service. IBM has a number of unique features, and lately has highlighted the sheer number of countries where it has data centers, a big concern for international organizations.

In fact, just about every manufacturer of data center hardware and software is trying to figure out how their offerings need to change in a world of cloud services, especially the enterprise infrastructure makers.

And that's not even counting the Software-as-a-Service (SaaS) players, such as Salesforce and Workday, which are offering compelling cloud applications.

Obviously, price is far from the only consideration in choosing what applications should go the cloud route, and which providers should handle them. It's not even the most important one. Far more important is understanding exactly what you want to go to the cloud, what's possible and likely, and how well the vendor's platform and plans fit in with your firms. It's almost as easy to get locked in to a cloud provider as to get locked into proprietary software. Other important topics include security, identity, flexibility, and the skills of your staff.

For the past several years, it's become common to consider cloud services as an alternative to on-premises applications and internal data centers. But with the new levels of pricing, controls, and options, it's now getting to the point where most organizations can assume cloud first, and only use internal applications and datacenters when they have a compelling reason to do so, at least for new applications. We're moving from a world where cloud services are no longer just an option, but are instead becoming the default.