When it comes to cloud computing, a lot changed in 2010. If the cloud were like an NFL quarterback, its status went from doubtful or even being on the injured reserve list to suddenly becoming everyone's favorite, smash-mouth player.
And In 2011, the cloud's definitely got game. Expect to see some clever handoffs and big gains in implementations over the next 12 months as interactions between the cloud and enterprise data center pick up. One thing that will emerge will be better management tools, tools that can address multiple hypervisor environments, like Novell's XENworks. Here's 5 predictions about what to expect in the coming year.
1. Payments In The Cloud
Nowhere is the changing environment more evident than in the possibility of Payment Card Industry (PCI) transactions, which weren't even in the cloud's playbook in 2009 or early 2010. They could materialize as a big factor in 2011. A transaction may still originate on company premises, but now it can be handed off to the public cloud, executed there, and still be in regulatory compliance.
On January 1, the Payment Card Industry Security Council implemented new regulations for PCI compliance, called PCI 2.0. While PCI 2.0 doesn't recognize cloud computing, it does recognize the secure operation of virtual machines. That takes PCI a giant step closer to the cloud.
In November, a technical subgroup of the council released a white paper, "PCI-Compliant Cloud Reference Architecture," that, while not endorsed by the council, clearly describes what's possible. It didn't take Amazon Web Services long to notice. On Dec. 7, AWS announced that a PCI auditor, IOActive, had certified EC2 for conducting Level 1 transactions under PCI regulations. No auditor had done that before.
In 2011, the PCI floodgates are likely to open and credit card data will flow into the cloud. The drafters of the next set of PCI regulations, version 3.0, will have to stand by over the next three years and watch as (hopefully) secure transactions are executed in many cloud settings. Instead of the cloud following the standard, it's likely the standard will be trying to catch up to the cloud by the time a PCI document uses that term for the first time. That would likely be in the fall 2013 timeframe.
Secure transactions in the cloud are good news to startups, which struggle to establish the separated infrastructure needed for compliance. Even larger companies may find they can further shrink their data centers by safely offloading the peak transaction load that materializes during their year-end selling season.
So in 2011, the cloud will become three things that critics said it never would: it will be certifiably secure for credit card transactions; many clouds will host not one but multiple virtual machine types in the same infrastructure; and the cloud will become easier to manage, no small feat considering the complexity of the first two.
2. Multiple Virtual Machines In A Single Cloud
All of this seemed extremely unlikely a short while ago. Could it have been only in September 2009 that Oracle CEO Larry Ellison was issuing another crowd-pleasing cloud bashing at the San Jose Churchchill Club? Now he's administering cloud-washes, where old products (Oracle Real Application Clusters and the database appliance) emerge out of the mist as new, shiny objects that are, themselves, in his words, part of "the cloud."
Multiple virtual machines in one cloud, really? That's not been the case so far. The cloud will become multi-hypervisor based because customers are adopting multi-hypervisor environments. As Windows Server 2008 Release 2 became a familiar part of the department or business unit, its free Hyper-V virtual machines have become part of the operating environment. Meanwhile, a whole section of the IT managed data center has been taken over by VMware and its ESX Server virtual machines.
As the market leader, Amazon required early cloud customers to come to it and submit their workloads as Amazon Machine Images, a proprietary version of an open source Xen virtual machine. Now it sees things differently. VMware seeded several cloud suppliers in 2009 and 2010 with its own management software, which optimizes handling ESX Server-based workloads. Thanks to that approach, cloud services are proliferating.
AT&T, with its synaptic compute cloud, and Verizon, with its Verizon Business Cloud service, are going to try to do an end-run around Amazon, the market leader, with more enterprise-oriented services based on hosting the virtual machines they use: VMware's.
Noting the change, Amazon on Dec. 16 announced in a blog, "Bring Your VMware Images to The Cloud." It's now offering a new tool, VM Import, that allows a VMware virtual machine to be recognized through its VM Import interface or through Amazon-supplied tools that give access to Elastic Compute Cloud (EC2) and Simple Storage Service (S3) through entries in the tool's command line.
Novell understands the enterprise is hosting multiple brands of virtual machines as well and upgraded its ZENworks management tool's ability to deal with multiple hypervisors in 2010. Whether it can push that approach forward after it's been acquired by Attachmate is another question. But the multi-hypervisor cat is out of the bag and all the major vendors are likely to have to support their rivals' virtual machines in the next generation of their management products.
3. Cloud Monitoring Leaps Forward
Monitoring will take more steps to overcoming a glaring weakness in cloud computing: an inability to see what your application is doing and how it's performing. Amazon enhanced its own CloudWatch in this direction in 2010; look for more of the same in 2011. Compuware, a former sophisticated software design tool supplier to automakers, shifted into the cloud market with CloudSleuth for long-distance application monitoring. VMware offers Hyperic. Watchers over the network segments between your workload and the cloud, such as Apparent Networks, will offer greater visibility into performance. But more needs to be done to directly spot problems in end-user response times.
4. Rise Of The Telecoms
The cloud formation is shifting in fundamental ways, too. Cloud computing infrastructure was first built by Google and Amazon.com, but the telecoms had similar experience in building highly reliable data centers as telecom switching centers. The cloud data centers innovated in shifting some of the responsibility for reliability and recoverability off the hardware and into the software running the server cluster. That made running applications remotely more viable. But in fundamental ways, the two had a great deal of similarity in efficient, automated operation. As noted before, AT&T and Verizon have moved decisively into becoming cloud suppliers. Look for more telecoms to do so in 2011.
One way they will circumvent Amazon is by being more enterprise oriented. It became evident earlier this year that Eli Lilly was in something of a dustup with Amazon over its service-level agreement (SLA). The issue was the weak remuneration that Amazon SLAs provide. Yes, Amazon will compensate you if your service goes down. But, basically, the SLA only allows a credit for an equal amount of computing time to replace the amount lost in an outage. That would be pennies compared to the value of business lost during downtime. All of which leaves room for maneuvering by telecoms, Terremark (a supplier of highly reliable services to the U.S. government), Rackspace, Savvis, and other suppliers.
5. Snappy Cloud Application Integration
As cloud environments proliferate and cloud access becomes an easier thing to manage, there will remain a nagging problem of getting one application to work with another, both in the public cloud, such as EC2, or the hybrid cloud, say between EC2 and a virtualized section of an enterprise data center. This was a big problem on premises; it gets bigger in the cloud.
New companies are likely to spring up to take advantage of the setting that the cloud itself offers. For example, SnapLogic, which just received $10 million in funding from Andreesen-Horowith and Floodgate in early December, puts its integration-building SnapLogic Server in the cloud, and urges customers to bring their integration needs to it.
If a user wants to connect his internal accounting system to, say, his Salesforce.com sales force automation system, SnapLogic provides (or he can build) a connector or "snap" between the accounting system and SnapLogic Server. The server has its own connectors to Salesforce.com, SAP, Oracle E-Business Suite, and other applications. At least half of the integration problem has been solved in an automated fashion; in many cases, the whole connection can be quickly established. If the data needs to be converted as it moves from one application to the other, SnapLogic Server recognizes that and automatically executes the reformatting.
The connection problem is the sleeper in the cloud. After you get there, how do you connect to standard package applications or even custom code without incurring a lot of work? SnapLogic, which recognizes the inherent value of the cloud setting, or other solutions like it, may be a big part of the answer.
