Two Ask TPM Questions: Real IBM i Clouds And Apple Buying IBM i
May 7, 2012 Hey, TPM
In AWS/400: Amazon Builds An AS/400-oid Cloud, you wrote:
I really do wish IBM would realize that IBM i is already a platform cloud of sorts, just waiting for it to be deployed on its SmartCloud so customers can get access to it and use it on a pay-per-use basis. If not, maybe we can get Amazon to roll in a few thousand Power System nodes and make use of DB2 Multisystem and other features to have Amazon build a truly scalable DB2 for i database service and build EC2-like images that work as RPG and COBOL compute engines that hit this database service.
What do you think is the possibility of that?
To answer your direct question, based on the excerpt you took from the final paragraph of the story, I think it is a very small possibility that Amazon would ever add Power Systems iron to its AWS cloud services unless it gets a lot of customer demand for the boxes, or better still, used the machines internally to run its vast e-tailing operation.
Amazon has created its own variants of the Linux operating system and the KVM server virtualization hypervisor as the foundation of the AWS cloud, and the entire AWS API stack hooks very tightly into this foundation. Moreover, all of the various database, middleware, and networking services that Amazon has created run atop this virtual computing foundation and link to each other through the API stack. While it would be theoretically possible to take the PowerVM hypervisor and IBM’s Systems Director or even Flex System Manager tools for the new PureSystems machines and link them into the Amazon APIs, this would be a very large undertaking and one that does not fit in with Amazon’s X86-only server strategy. And doing so would not get you what I am talking about anyway.
You have to read the two final sentences in the story to see how we can get to true IBM i clouds:
Cloud computing starts out as hosting and evolves into something more flexible and malleable, and as I learned in biology, ontogeny recapitulates phylogeny. Which basically means the evolution of the individual tends to go through the evolution of the species as a whole, in a very rough sense. You don’t get to skip steps. IBM i shops are just being presented with affordable virtual system slices from the likes of Abacus, Connectria, and others, and given the fact that most IBM i shops don’t have high availability clustering and replication software installed and are running business-critical apps, this is often where they start because this is the real worry. But eventually, they will want daily or hourly metering, automatic scaling, database service and clustering, various network and acceleration services, and big data munching on their information stored in DB2 for i database. And I think it will be service providers who figure out how to do this ahead of IBM or Amazon for the IBM i platform.
I think the first thing IBM will need to do to foster this is to give service providers cheap Power 720 machines like the PowerLinux boxes announced two weeks ago, enable IBM i, DB2 Multisystem, parallel SQL execution across nodes, and other features on these machines to allow for clustering of these cheap nodes to run middleware and databases. The live migration of LPARs, which comes out with Tech Refresh 4 later this month for IBM i 7.1 is a key cloudy component, too. These features would result in a truly cloudy IBM i setup. It is just too hard to make a Power 795 make economic sense as a multi-tenant cloud, as impressive as it would be to use a giant shared memory system instead of a cluster of 16 machines to offer 256 Power7 cores for workloads to play on. The trick in the modern hyperscale world is to write scale out software that can span thousands of cores if necessary, but quickly collapse down to two cores if that is all you need.
Apple to buy “i” division from IBM? How about that for an idea?
I’ve suggested to my local IBMers that this would be a great thing for the platform. First, remove it from the IBM mantra of “we don’t have our OSes compete against each other, but we’ll advertise all but i,” and second, the marketing opportunities that this would create.
I’ve already chastised IBM for letting the “iCloud” trademark go unregistered.
Over the years, myself and others here at IT Jungle have toyed with all kinds of possibilities, even suggesting way before the iPod, iPhone, and iPad hit the market that IBM should buy Apple and ditch its own PC biz. (Big Blue half listened.) We could imagine the MacOS as the front end GUI on Power Systems iron, and it seemed like a good idea to us.
We also suggested that Apple should have worked with IBM to port its MacOS operating system to IBM’s Power Systems iron since its own PowerPC-based Xserve servers and storage were pretty uninspiring. And rather than listen, Apple ditched the server business entirely instead.
IT vendors are just like parents and children alike: they don’t listen.
There are many parallels between MacOS and OS/400-IBM i and Apple and the Rochester part of Big Blue, no doubt about it. And there would be some very interesting things that Apple might be able to do with IBM i. But I am sorry to say that Apple just doesn’t seem to be interested in back-end infrastructure.
It is hard to blame Apple. It is sitting on $100 billion in cash from its consumer computer and smartphone boom and it has built a pretty impressive electronic media empire. By any measure, Apple has succeeded to its wildest expectations and it just does not need a formal IT infrastructure business. There may come a day, however, when Apple does need to build servers to more cheaply support its own vast cloud business, and it would not be surprising to see Apple pull an Amazon and start selling excess capacity on its own cloud to outsiders to run their own apps.
For Cloudy IBM i:
For Apple buying IBM i: