We’ve already explored the technical similarities and differences between traditional mainframe and x86 virtualization architectures in a two-part feature article on SearchDataCenter.com. This series of podcasts focuses on these technologies and how they relate to the evolution of cloud computing, as well as predictions for the future of the market. In previous episodes, we took a critical look at the “software mainframe” analogy, and discussed future predictions. In this podcast, our experts offer cautionary statements to go along with their forward-looking predictions.
Beth Pariseau, senior news writer, discusses these topics with Wayne Kernochan, president of Infostructure Associates LLC, an affiliate of Valley View Ventures Inc., and David Floyer, chief technology officer for Wikibon.
You can also listen to a podcast of this Q&A here:
Pariseau: Just as companies giving forward-looking statements also offer cautionary statements, I’m wondering what your cautionary statements would be? What circumstances or issues might challenge your outlook on whether x86 or mainframe architectures will dominate in the next 10 years, and what do you mean when you talk about vendors needing to get their act together?
Floyer: Let me have a go at several weak points in the drive toward the software mainframe. One potential weak point is Intel – it has a lot on its plate and has had its fair share of hiccups – Itanium, for example, being one of the most painful in its history. If it does not execute on its plans and find problems within its plans, it could have a profound impact on the predominance of x86 architectures.
A more likely problem is that the attempts to put in all of the layers of security, availability, etc. that need to be in the platform under a virtualization umbrella will slowly grind to an agonizing halt. Software architectures can get bound up with themselves. They can have common choke points, and designing this sort of architecture and making sure that it will work for both high performance and for the lower end of the marketplace is a big challenge. Therefore, the time scale for implementation of security and availability, and the management aspects of that, may well be longer than the two- to three-year horizon in Paul Maritz’s vision [when he coined the “software mainframe” term]. We’re talking about a five- to 10-year horizon, but it could well push out beyond that.
The real driver [for x86 architecture] is social media, new startups in [the database market] and big data. There will be multiple solutions on multiple virtualization platforms that will come out of this effort, not a single software mainframe as Paul Maritz would like to envisage.
Kernochan: First, David has already given half my answer for me. Let’s say, in three years, NAND storage is out in the market and thriving; I will very happily eat my words.
The second thing that would cast doubt on my viewpoint is if the virtualization vendors on the x86 side move faster than I’ve seen them move before. Obviously, IBM has not been slow to upgrade, extend and make common its own virtualization solution, and all those moves have cemented the mainframe’s place in a virtual environment. Where VMware has certainly been a leader in the past in that area, if the technology in general moves faster than it has before on things like robustness, security and so on, then that will crowd IBM and the mainframe in the here and now.
The third thing that I don’t think will actually be the main cause of any failure of my prediction, but it’s something I think folks should keep an eye out for, is that up until now, IBM, and the mainframe folks in particular, have not fully appreciated Windows on x86. There are ways of getting quite integrated with Windows workloads under a mainframe-dominated scale-up type of system, and IBM has effectively chosen not to pursue them.
[IBM has] indicated that it’s beginning to change that – it’s looking to the new zBX to provide Windows support on the new [zBX] by the end of this year. But if IBM continues to lag completely, it’s basically walking away from public cloud environments, because there is an awful lot of Windows on x86, and you’re not going to put it in a machine that doesn’t have the flexibility to handle a major part of your workloads. In turn, the public cloud would be much less mainframe-based, and IBM would have much less stake in the hybrid systems of the future.
This was first published in April 2011