2 Comments

Summary:

I made a lot of predictions about cloud computing and the general IT infrastructure space heading into 2011, and I impressed myself with my skills of prognostication. Of course, it’s possible I’m just grading myself too generously, so I’ll let readers be the judges.

crystal ball

I made a lot of predictions about cloud computing and the general IT infrastructure space heading into 2011, and I impressed myself with my skills of prognostication. Of course, some might have been no-brainers, and it’s possible I’m just grading myself too generously, so I’ll let readers be the judges.

For your assessment, here’s what I predicted (the full predictions are all available on GigaOM Pro, subscription required) and a synopsis of what actually happened. Comment away.

5 trends not to expect

Ubiquitous cloud adoption. This might be one of those no-brainers. We certainly heard a lot more about large enterprises experimenting with the cloud this year, but we’re still nowhere near ubiquity, especially when it comes to serious applications. I’d argue we’re getting there, however, as Amazon Web Services’ incessant enterprise push makes it more appealing and newcomers such as Virtustream successfully court even ERP applications.

Amazon will announce its AWS revenues. This didn’t happen, but it didn’t stop the speculation. When I looked at AWS’s third-quarter numbers, it looked like the business might be at a billion-dollar run rate. That might not be so impressive, however, if analyst predictions of a mere 4-percent net profit margin are accurate. Maybe Amazon should let us know what’s really going on.

Apple will tell us anything about its data center operations. Oh, it showed some pictures that got the blogosphere talking, but Apple didn’t tell us anything. What we know: There’s a big, expensive data center in North Carolina, probably running a Hadoop cluster. What we think we know: iIt’s full of Teradata appliances, Oracle Exadata appliances and HP servers.

Legitimate progress on clean IT. There has been nothing that I can tell, unless you consider the truce between Facebook and Greenpeace legitimate. We saw more funding for data center energy-management software and some innovative practices from Google, Facebook and others, but surveys show clean technology is still out of reach for most data center operators. Solyndra’s collapse didn’t help the cause, either for cheap solar film or around clean-energy policy.

The demise of Intel. This was perhaps another no-brainer, although 2010 was rife with speculation about Intel’s absence in the mobile space and the potential for ARM-based processors to disrupt its data center domination. Well, record quarters and a willingness to embrace low-power servers suggest Intel is alive and well in the data center, and it probably has something up its sleeve around mobile devices.

8 companies poised to make headlines

Calxeda. This one is a toss-up, but I think Calxeda’s partnership with HP  is a big deal. Servers built using Calxeda’s uber-efficient ARM-based processors won’t be available for at least another few months, but having the world’s largest server vendor pushing them is a definite coup.

Cisco. Cisco made headlines, alright, but probably not for the right reasons. Rather than talk about its grand play in cloud computing or a network revolution, Cisco’s year was highlighted by talk of mass layoffs and a realignment of corporate strategy. Buried in the bad news, however, was the shining star that is UCS server sales and Cisco’s growing involvement with OpenStack.

CouchOne. CouchOne certainly made headlines when it merged with Membase in February, regardless the reasons behind the deal. Since then, Couchbase has raised a lot of money, released some significant products and, I’m told, is raking in revenue beyond that even of NoSQL mindshare leader 10gen and its MongoDB-centric business.

Salesforce.com. This one might have been too easy, as everything Salesforce.com does makes headlines, it seems. The company itself made a handful of acquisitions, but I’d argue that new Salesforce.com division Heroku kept leading the Platform-as-a-Service charge by improving its service, adding new languages and launching a database service. It also got a potentially lucrative hook into the Facebook developer ecosystem.

Joyent. Joyent’s year might not seem too impressive until you look under the covers a bit. The company says its private-cloud software business is doing very well, primarily among service providers. Joyent also redesigned its public cloud to look a lot more like AWS on the surface and with pricing, but, it claims, with much better performance. And as Node.js takes off, community leader Joyent is poised to cash in.

OpenStack. Internap already running OpenStack software; HP is running an OpenStack-based cloud in preview mode; and Dell is planning one in 2012. Former NASA cloud leaders Chris Kemp and Joshua McKenty (see disclosure) are both launching their own OpenStack-based startups. That’s three major code releases. I’d say OpenStack did alright for itself in 2011.

Red Hat. It’s now officially the first billion-dollar, open-source company. Enough said. It also spearheaded the Open Virtualization Alliance to push for products using the KVM hypervisor, as Red Hat’s own virtualization software does. HP, Intel and IBM are on board.

Twitter. It was bittersweet, but Twitter certainly made infrastructure headlines. Some were about its data center woes — specifically, where it’s now located and how many moves did it take to get there. However, it also bought real-time analytics startups Julpan and BackType, and open-sourced the Storm real-time data-processing software that came along with the latter.

5 things that will happen

More specialized clouds and cloud services. I don’t know that we actually saw too many specialized clouds emerge (unless you count enterprise-application-focused clouds such as Virtustream and Tier3), but we definitely saw more specialized cloud services. Cloud-based security and data services, in particular, began either popping up or getting at least getting more attention.

Infrastructure startups will get bough.t OK, so I picked Cloudera, Eucalyptus and Engine Yard as candidates and was wrong on all three. Who got bought: CouchOne, Cloud.com, Endeca, Orchestra (by Engine Yard), OpSource, Gluster, Vertica, Aster Data Systems, Autonomy, Cotendo, Zeus, Savvis, Terremark, Terracotta, Platform Computing, IO Turbine, Force10 Networks, Anso Labs, and NaviSite.

The fusion of IaaS and PaaS will continue. It’s hard to argue this didn’t happen, especially if we look at the emergence of new platforms such as DotCloud, the way Heroku has matured and Amazon’s Elastic Beanstalk offering. VMware’s Cloud Foundry project, although very PaaS-centric right now, could help seal the deal assuming it eventually merges more tightly with VMware’s core infrastructure business.

Low-power processors will have their day. They’re not commonplace yet, but they’ve had their day, for sure: HP partnered with Calxeda; Intel built special processors for SeaMicro and announced its own low-power Xeon roadmap; and eHarmony is running a production Hadoop cluster on SeaMicro. At the very least, low-power processors have proved they have value, even if not for every workload.

A cloud provider will end up in court. Unless I missed something, this didn’t happen — but it certainly could have. Frankly, I’m surprised the AWS outage in April hasn’t resulted in any lawsuits (at least that have been made public), but maybe that’s just because the most-affected parties were mostly small. This will happen sooner or later though, but maybe it will take a data breach or some sort of Fourth Amendment issue to make it happen.

Disclosure: Piston is backed by True Ventures, a venture capital firm that is an investor in the parent company of this blog, Giga Omni Media. Om Malik, founder of Giga Omni Media, is also a venture partner at True.

Feature image courtesy of Flickr user Bitterjug.

You’re subscribed! If you like, you can update your settings

  1. it’s hard to understand how AWS could not be running a pretty comfortable profit margin. it’s easy to look at their prices – say, $3k/year for a large on-demand instance, which is about half a core of a generic 12-core server which would list for about $3k. power is going to run about 10% of that per year, and their PUE can’t possibly be worse than 1.3 or so. still need to account for space, maintenance, UPS, network. but that’s still a pretty astounding markup. and this isn’t even assuming any overcommitment (euphemistically called “consolidation”).

  2. Aditya Jayaram Friday, December 30, 2011

    Great roundup of what we saw of the cloud in 2011.With businesses getting to grips with the cloud on an experimental basis, adoption of the cloud is set to accelerate in 2012.Global accessibility and interoperability will be the key factors driving cloud adoption.Just watched an impressive video on cloud computing @http://bit.ly/uMMik3

Comments have been disabled for this post