Cloudy, With a Chance of Mainframe
IBM Looks to Carve Its Own System z Slice of the Cloud Computing Pie
7/1/2011 8:25:58 AM |
By Ryan Rhodes
"Cloud computing" is one of the hottest terms being bandied about today. Everyone seems to be talking about "The Cloud" and how it's revolutionizing the way companies and individuals engage in business and recreation online.
While cloud computing has certainly generated its share of buzz, defining exactly what it is varies considerably depending on the audience. At its most basic, cloud computing refers to the IT model of delivering and procuring hosted services via the Internet, including Software-as-a-Service (SaaS), Platform-as-a-Service (PaaS), Business-Process-as-a-Service (BPaaS) and Infrastructure-as-a-Service (IaaS). Within that nebulous definition, however, exists considerable room for refining and otherwise tweaking what cloud computing means for each unique implementation.
Amazon and Google maintain the most popular public cloud offerings, while private cloud computing providers—including IBM—provide clients with more overall control and management of their chosen proprietary offerings, from selection to implementation to management to eventual contract termination. For IBM, entering the world of cloud computing—particularly with the mainframe in mind—has required an even more nuanced breakdown as to what cloud computing is to IBM and what it means to its established and potential clients.
"The way I see it is to basically say 'Cloud is really broken down into six steps,' " explains Bill Reeder, IBM cloud computing and IT optimization sales leader for System z. "First, it's figuring out what service or services we're going to provide and, second, figuring out the necessary steps to create and automate as much as possible to provide the service and services and then, three, contracting for that, whether it's internally, externally or whatever. Step four is deploying what it was you just contracted, and step five is the management, care, feeding, service-level agreements (SLAs), disaster recovery and the rest of the operational end of it all. Finally, step six is decommissioning it all at the end of whatever cycle you may be operating within."
"By going back to step one, we can better ask, 'What service do you want to deploy?' " says Reeder. "The service could be a server, or a function like Apache Web server, or it can be a very complex solution like a whole ERP suite—or anything in between. If you start thinking along those lines, you start looking at the overall cloud possibilities, which can be anything from simple infrastructure to databases to messaging and so on."
As previously mentioned, cloud computing can include any or all of the following: SaaS, PaaS, BPaaS IaaS, which can cover anything as simple as data storage to comparatively complex business operations, analytics, monitoring, etc. According to Reeder, the System z platform can deliver all of that and more, but the longer term goal is to provide more of a hybrid cloud solution.
"Right now, most of the cloud offerings with System z are just that—only System z—meaning customers can only really leverage z/VM and Linux on the platform," says Reeder. "With a hybrid cloud, customers can also leverage things like the Tivoli product suite to start managing into other environments. So a cloud can be built to optimize some of the workloads on the IBM zEnterprise BladeCenter Extension (zBX) on either an x86 Power blade or a DataPower blade, or maybe a whole subset of rendering software sitting on an array of POWER processors—all that could be managed from Tivoli Server and Automation (TSAM)."
At this point, Reeder is quick to establish the cloud crystal ball begins to relay varied messages as to the future direction of System z and its eventual role in the cloud computing landscape, because some technology pieces are currently more mature than others. While there are fundamentally three current cloud deployment models in the works for System z, they are not working collaboratively quite yet.
For example, according to Reeder, a TSAM configuration can deploy to Linux on System z, Power, Sun, HP, VMware, etc., which represents one model. Alternatively, the zEnterprise Unified Resource Manager (zManager) model is designed to manage resources and deploy workloads into just System z and zBX environments, but it doesn't yet communicate to TSAM. IBM Director represents yet another deployment model option.
"With all of the hypervisors and controls we're talking about when it comes to System z and the hybrid cloud models, it would be nice to have a set of program interfaces where they all work and talk together," says Reeder. "The only problem with that whole last sentence was 'it would be nice to have.' We don't quite have that yet, but that's what we're working towards. In truth, no one else has it either, except within a monolithic structure."
Eventually, once IBM clears the communication and interoperability hurdles, the System z platform could become the cloud solution platform of choice for enterprise-level organizations looking to leverage the convenience of the cloud along with the security, reliability, availability, scalability, high CPU utilization and overall efficiency that are hallmarks of the mainframe. With the added convenience of eventually having a hybrid cloud computing model at their disposal, customers may indeed decide the mainframe cloud offers the best available solution for their unique business requirements.
"Since no customer I'm aware of uses only one architecture and one operating system, our hybrid approach should provide a real business advantage," says Reeder. "For example, no one is running all SPARC or Oracle products only, or even only all mainframe for that matter; there are always still some other things. The ability to manage the other things and your core business from one place, to me, is really quite compelling."