The all-in cloud architecture of tomorrow

Not all of us pundits at InfoWorld agree with each other, but regular readers can put together a kind of consensus architecture. That consensus represents the state of our industry and the thought leadership that is working to determine its near-future direction. As we approach the end of the year and finish off the first dirty dozen of the 2000s, it’s useful to consider what lies ahead if we actually do everything we talk about. For me, this extends to a New Year’s resolution: I plan to practice what we preach in 2013 and move to an all-in cloud architecture. Here’s what that will entail.

Bring your own device

Frankly, BYOD is the prescription I have the most trouble with — and the one that my own company is the most conservative toward. There are various legal, marketing, and staff development reasons for this. Nonetheless, BYOD a great positive force for our industry.

[ Also on InfoWorld: 10 things you needn’t worry about in 2013 | Learn how to work smarter, not harder with InfoWorld’s roundup of all the tips and trends programmers need to know in the Developers’ Survival Guide. Download the PDF today! | Keep up with the latest developer news with InfoWorld’s Developer World newsletter. ]

First of all, it means you must pursue a standards-based architecture: no lip service, no “requires Internet Explorer,” no “visit the Apple Store,” but go full-on HTTP-based architecture. This changes everything from your security architecture to the way your organization picks applications.

Software as a service When it’s pure SaaS, there’s nothing to install. There’s nothing to maintain. Support, infrastructure, storage, scale, and reliability are all bundled as one package. And of course, SaaS is priced as a service, typically per month/per user.

This is the model that failed toward the end of the dot-com boom in the ’90s and that many doubted would come back to life. Now, even Microsoft is in the game with Office 365. Our company uses Google Apps for nearly everything. I’ve worked with pretty big companies recently that have also gone the Google route for both their office productivity suite and email. And many, many companies are addicted to the Salesforce suite. Salesforce has been the poster child for SaaS, and its success has emboldened the market and made believers out of some of the more cautious and skeptical CIOs and CTOs.

On its face, SaaS can be expensive — and you must be careful that you have an exit strategy in the event the vendor raises the price or vaporizes. But when you take into account total cost of ownership, SaaS is almost always worth it against what you’d pay to install and support a similar product internally.

Platform as a service With conventional application development, you stand up a bunch of servers and dedicate a team to supporting the whole endeavor. Virtualization has eased that burden somewhat. IaaS (infrastructure as a service) makes provisioning and reconfiguring environments even easier, though I still view IaaS glorified virtual hosting (flame away, I come armed).

Now we have PaaS (platform as a service), which for app dev is where you go to fully reap the financial and technical reward of the public cloud. PaaS means just deploying your application and letting it rip. Scaling it, making it highly available (caveat: you have to write it correctly), the infrastructure, the database, and all the fixings are taken care of for you. For Java, as an example, this means you deploy a WAR file, give it your scalability budget, and walk away.

NoSQL and big data Hoarders is more than a reality TV show — it’s the data strategy for most modern corporations. The RDBMS has been a great unifying force, the standard no one dared challenge. But it’s becoming antiquated, and we used it for way too much. After years of struggling with database tuning and huge license fees to scale a technology that was never designed for massive scale, the industry is addressing the problem.

It seems clear that most operational systems will use a document database like Couchbase 2.0 or MongoDB. There will be more room for utility databases like graph database Neo4j for specific purposes, but they probably won’t be your primary data store. It also seems clear that on the analytical side, MapReduce frameworks like Hadoop may still be waiting for a killer product to break away from the pack.

Your network In our all-in cloud architecture of the future, what you don’t have is as important as what you do have. You don’t have company-installed and company-supported desktop applications. You don’t have file servers (because SaaS applications take the place of them), and you don’t have Windows networking infrastructure. Beyond the basic switches and routers, as well as whatever you need to get people to the Internet, your network is naked.

Your people This move could send highly qualified system administrators to work as baristas — or worse, at the Genius Bar. There is as much of a labor consolidation here as a hardware and software consolidation. This move is about our industry turning on itself and providing productivity gains (read: people get furloughed and the rest do more work). Someone still has to figure out why the dang wireless router keeps skipping out, but beyond that the skills of the IT administrator will change.

The all-in cloud architecture has lots of moving parts, particularly the portals that get your data or your users between the clouds. Your skilled administrators will speak a language of configuration based on SAML, Oauth, and autoscale. They’ll know enough about the database to be dangerous and diagnose problems, but to some degree the rise of the cloud-deployed NoSQL and big data means that DBAs will be brought back kicking and screaming from their graves. That DBA may resist the already decade-long move away from PL/SQL stored procedures, but the rise of NoSQL and big data may actually revive a position that has long been in decline.

The costs The all-in cloud architecture will reduce our level of control. In some way that is the point; in other ways it will be frustrating. It also requires high-bandwidth, more reliable networks to work. This will increase the digital divide not only among people but among companies. Outages will be more catastrophic when they happen and affect more companies at once. These will make headlines, but overall, reliability will be higher. It will be like when a jetliner goes down: It makes the national news, whereas thousands of fatal car wrecks don’t. There will also be bad actors in the cloud space, thieves who convince you to migrate, jack up the price, and never let you leave. Freedom to exit will become the most important freedom, if it isn’t already.

The benefits The all-in architecture will be cheaper and more reliable. It will scale better, offer high availability, and require less maintenance. On the whole this will represent economic consolidation in the tech industry. Hardware manufacturers won’t have a good time of it, but everything else will be more efficient. That might sound bad for the tech industry, but just like energy efficiency doesn’t help the environment though it increases the demand for energy, efficiency in the tech sector will increase demand for technology.

What we don’t know yet As I mentioned in a previous post, we still don’t have a standardized service directory. We also lack an Active Directory killer. Sure, there are several players, but there are no clear winners yet.

We also don’t know when and how long this transition to the cloud will take — it just feels like we are approaching a tipping point, living in a moment before everything changes. In 2013, I plan to go all in. Care to join me?

This article, “The all-in cloud architecture of tomorrow,” was originally published at InfoWorld.com. Read more of Andrew C. Oliver’s Strategic Developer blog, and keep up on the latest developments in application development at InfoWorld.com For the latest business technology news, follow InfoWorld.com on Twitter.