June 6, 2011 Timothy Prickett Morgan
It is perhaps fitting to discuss the future of cloud computing on the IBM i platform on the same day that IT and consumer electronics darling Apple is expected to launch iCloud, its fourth iteration of online services for its various iMac, iPhone, iPod, and iPad devices. If anyone has dibs on the iCloud name, you would think it would be IBM with its iSeries, System i, and Power Systems-IBM i platforms. But Apple ponied up the $4.5 million to buy the iCloud name from its owner and beat IBM to the punch.
Exactly what iCloud is–and isn’t–remains cloaked in mystery, of course, because Apple locks things down pretty tightly. But the chatter out there on the Intertubes where Apple has found its resurrection and salvation is that iCloud will be a set of data storage services for its various styles of computers and operation systems being delivered from its brand-spanking-new, $1 billion data center in North Carolina, probably with a hub in California and maybe one each in Asia and Europe at some point.
Apple’s iCloud is not, however, going to be what IT people mean when they say “cloud,” which is a set of virtualized server instances suitable for encapsulating and running business applications. Apple has shown no inclination to get into corporate applications, and there is no reason to believe–particularly after it has killed off its server business earlier this year–that Apple has suddenly taken a shining to ERP and other kinds of back-office software. The only apps Apple care about are ones that cost a few bucks and drive people to buy iPhones and iPads, and considering that Apple is now throwing off more profits than IBM these days, it is hard to argue that Steve Jobs is wrong about selling machines and apps to fanbois and fangirlz instead of trying to take on server makers and Amazon‘s EC2 compute cloud and related services.
IBM, as it turns out, has built its SmartClouds for public use and CloudBursts for building private clouds, based on X64 and Power processors in the latter case and someday soon both architectures will represented in the former. But when it comes to the IBM i platform, Big Blue is thinking about clouds in much the same way Apple seems to be doing with iCloud–as a storage platform, not an execution environment. This may seem peculiar to you, given how many remote storage and disaster recovery services have been launched in the past decade as storage and telecom costs have come down, and how many of them have failed–including EMC’s own Atmos storage cloud. (Let’s face it, if EMC can’t build an enterprise storage cloud at cost, then who can?)
But remote storage and disaster recovery is the first step in the IBM i Cloud strategy, according to the company’s top Power Systems brass. And this is what customers are asking for, apparently.
“First of all, I think there is demand for i clouds,” explained Tom Rosamilia, general manager of the Power Systems and System z division at IBM, in a meeting last month at the COMMON midrange user group meeting in Minneapolis. “I am seeing it today and I just came out of a meeting with users who are very interested in i clouds. But I would say that it is less about what I would view as a traditional on premises private cloud and is focused more on backup capabilities. That didn’t get into the dynamic mobility of anywhere, anytime, anyhow. In terms of mobility, that’s far more likely to wait until the next release for us to get to.”
This may be surprising to you, but maybe not, since you probably are not envisioning moving your OS/400 and i applications outside of your data center anyway.
It was also something of a surprise to Colin Parris, vice president and business line manager for the Power Systems business, who reports directly to Rosamilia and who basically runs the Power server biz.
“I went with the view that every client wants a cloud,” Parris explained. “I said that i clients would love a cloud, and I got slapped around a little bit. So let me tell you what they tell me they would like. They would like the DR focus first because they have to do DR, and in many cases they are small companies and they do not have another site. The customers tell me to look at the long tail of applications that they have. They have some applications that they are using infrequently and they do not want to invest in, and mainly because most of the time they have a very small IT staff. So if they can identify what those applications are and use them in the cloud, they will be OK because they are not mission-critical. They use them infrequently. Now, as they get more comfortable with those applications in the cloud, then they might be willing to think about moving some other applications across. But they will never want to move most of their big, mission-critical applications. They don’t think that way. So their view has been to start on the DR, and then to the other pieces in the long tail.”
The other thing that IBM i shops are telling the company they want is remote administration of systems and applications in an effort to cut costs. They will try a disaster recovery service hosted out on the cloud, and then maybe a hosted app or two of non-critical applications, and then perhaps some remote administration, gradually easing themselves into the cloudy world. This is why IBM seems to be in no hurry to get the same kinds of live migration and related workload partitioning capabilities ported to IBM i platforms.
“This is the way we see it evolving right now,” said Parris. “I am going to be guided by the clients right now because I think the IBM i client actually does understand what they need to do. Once they do more and more modernization, it becomes very interesting. Now you can talk about moving things back and forth and migrating things freely between the clouds. But we have a long way to go from the environment they are accustomed to.”
Rosamilia adds that another area where cloud might take off on the IBM i platform is application development and testing, which makes a certain amount of sense. “They can do app dev in the cloud, but they want their runtime to be in their four walls.”
I am often out there on the bleeding edge, arguing that the OS/400, i5/OS, and IBM i platform should be on the bleeding edge of any new technology, because it is that edge that gave the AS/400 its market muscle two decades ago. But at this point, the Power Systems-IBM i business is not trying to take on the world, but rather keep its paying customers happy with what they ask for. I want a cloud that is essentially EC2 supporting IBM i applications, internally and externally, and told Rosamilia and Parris as much on your behalf.
“You are talking about a world of truly dynamic provisioning, where you are going to ebb and flow as you run multi-tenant and supporting multiple lines of business on a highly virtualized, workload-isolated system,” said Rosamilia. “That, to me, is a great challenge for a runtime cloud. That is not what our IBM i clients are looking for right now. That is what some of our other clients are looking for right now.”
And so, don’t hold your breath waiting for SmartClouds or CloudBursts running IBM i, but maybe pencil them into your calendar for 2012 or 2013.