IBM i Wish List For 2015
January 19, 2015 Timothy Prickett Morgan
It is the beginning of a new year, and this is the appropriate time to ponder the things we would like to see happen in the IBM i community over the coming year. As I have said many times, the only way you ever get anything in this world is to ask for it, so I took a few moments to come up with a list of things that I would like to see IBM do in the coming year to help support and extend the IBM i community. Please let us know what you would like to see happen this year, too, and we will let Big Blue know what you are thinking.
The one thing that most IBM i shops are always asking for is more marketing and sales support for the platform and the resellers and software developers who push the iron and the applications that run on top of it. With IBM keen on peddling its own platforms now and not positioning itself as a generic supplier of servers, storage, and systems software, there is a much better chance that IBM will sell its platforms as platforms. I expect for Big Blue to push its System z platform pretty hard, particularly with a new System z13 machine entering the field. IBM has been promoting Linux on Power like crazy, too, as it tries to create growth for the Power Systems line, and it would be wise to not forget about either IBM i or AIX on Power. There are applications and customers that are suited well to AIX and IBM i. I think it is highly unlikely that IBM has an excess of marketing money laying around, however. Much of the marketing money in the IT sector comes from giants like Intel and Microsoft, and Big Blue was used to getting those co-marketing funds much as the IBM i community is used to getting help for advertising and marketing from IBM itself. It is not clear to me what funds IBM has set aside for marketing and sales for its platforms, but presumably at the front-end of a mainframe upswing and the ramping of Power8, the company has funds to help promote these platforms.
I think, if the IBM reorganization pans out the way I expect it will and that I reported on in last week’s issue of The Four Hundred, I already am going to get one of my long-time wishes. And that is for IBM to start talking about its systems business as a whole instead of carving it all up into hardware, software, and services sales. IBM is a systems company first and foremost, and the shift to separate groups for servers and software and services that the company did 20 years ago is one that perhaps made sense at the time but which makes no sense with the product lineup IBM has today. (Back then, IBM wanted to show its prowess in servers and software and that both were independent of its own hardware groups.) I have always contended that the bulk of IBM’s revenues and profits come from customers who buy an IBM stack, pretty much top to bottom but not always because IBM did not create Linux or Windows. Therefore, IBM should show the strength of the System z and Power Systems businesses as they are actually consumed by companies, as whole units with servers, storage, operating systems, middleware, and databases all stacked up to support applications that are either homegrown or developed by third parties.
That’s two IBM i wishes for 2015. My third wish is something I have been thinking about since IBM sold off its chip manufacturing business to GlobalFoundries last year, and indeed it is something that IBM could have and should have been doing all along, and that is putting out a rolling five-year roadmap for the Power chips. Intel does not do this publicly, and if IBM wants to demonstrate its commitment to the Power chip, it needs to lay out the plan, openly and publicly, so customers who are thinking of adopting the Power platform can show that it has a future beyond the current generation. Everyone knows that chip roadmaps change as process technologies and feature sets evolve, so it is not like IBM is committing to a precise launch schedule. But as I have pointed out in the past, when Oracle bought Sun Microsystems six years ago, one of the first things Oracle did was put out a five-year roadmap for its midrange and high-end Sparc chips. And to its credit, Oracle pretty much hit all of the Sparc towns on the roadmap on the way to the end of 2014. Now, people believe Larry Ellison when he says Oracle wants to build its own hardware. Incidentally, I was talking to Brad McCredie recently, who heads up IBM’s contributions to the OpenPower Foundation, and he says that Big Blue is considering putting together a future Power chip roadmap in conjunction with its OpenPower chip development partners. By the way, Intel does not provide such a long-term roadmap, and if it did, there would be a lot of churn because, to its credit, Intel’s Atom and Xeon chip designs are reacting to changing market conditions as well as the realities of the engineering and manufacturing challenges all chip makers face as their chips get more complex and wafer-baking processes get more cranky.
For my fourth wish for the IBM i platform for 2015 is what I think is probably a relatively simple one. I want to IBM to create a variant of the Portable Application Solution Environment, or PASE, runtime environment that runs Linux applications within IBM i. The PASE tool, which dates back 15 years ago to OS/400 V5, is an AIX environment that is embedded inside of OS/400, i5/OS, and IBM i that can run AIX applications directly on Power iron (or rather, its SLIC microcode). Many elements of the IBM i operating system, including the Java stack and the TCP/IP networking stack, actually run inside PASE and are not, strictly speaking, native applications for IBM i. But as far as IBM i is concerned, they can be run and managed like elements of IBM i or applications that run atop them. While IBM has been promoting Linux partitions on Power with Red Hat, SUSE Linux, or Canonical Linux distributions, with either PowerVM or PowerKVM as the hypervisor, being able to run Linux applications in PASE might be more useful in a lot of cases. IBM could make a PASE environment for the major Linuxes above, with varying runtimes tuned for the Power architecture as it does for the full-blown Enterprise Linux, Enterprise Server, and Ubuntu Server Linuxes from the vendors above. There is no reason why these could not be mixed and matched on a single system, allowing for a slew of Linux apps to run on the box without having to install partitions or full-blown Linux operating systems. You could call it the Linux Application Solution Environment, and no, it would not be QuickTransit emulation software running Linux apps atop AIX or IBM i, but the actual Linux runtime in a sandbox.
This is certainly easier than the fifth wish I have, and that is for a similar Windows runtime environment on Power. But as we have pointed out many times, without Microsoft supporting such a runtime environment on Power chips, it would be very difficult to run the key Microsoft system software on Power. While most Power Systems shops have Windows machinery wrapping around their central databases and applications, it is debatable if they would be willing to pull those Windows programs onto Power platforms if a Windows option was available. If the economics made sense and the integration was tighter, faster, and easier, then a technical case for a Windows Application Solution Environment, or WASE, could be made. If running Microsoft’s own code is not possible in such a WASE tool, then it should be possible to take the open source Mono.NET runtime environment, or the recently open sourced .NET Core runtime from Microsoft, and create a runtime for Windows applications. Many of us have been advocating for this for a long time. Run the Windows applications on Power and don’t worry about running Windows itself on Power. (Of course, such a strategy would depend on how much of the Windows Server and related systems software stack is used by an application.) Microsoft is supporting .NET running atop Linux and Mac OS, so it is even possible to conceive of .NET running inside of the LASE tool outlined above.
Many of us would love to see a full-blown Windows Server 2012 running inside of partitions on Power Systems machines. This is technically possible, since the Windows kernel was already ported to Power chips for the Xbox game console when it was based on Power rather than X86 processors. If this could come to pass–and there is no indication that it will–porting Windows to the new PowerKVM hypervisor instead of IBM-created PowerVM hypervisor is probably the right way to go. Windows already runs–and runs pretty well–atop KVM.
As we have reported already, IBM is working with its partners in the OpenPower Foundation to create machines based on the Power8 processors that do not come from Big Blue itself but, in some cases, will be sold by Big Blue. The company has hinted that these third-party Power8 systems will focus on Linux and will not be able to run either AIX or IBM i. For my seventh wish for 2015, I think that IBM ought to let its partners and its customers decide if they want to run AIX or IBM i on a specific system. All that IBM needs to do is offer a fair and equivalent price for its AIX and IBM i stacks on these third-party machines. This increases the support matrix, to be sure, and adds costs. Which is why I think it will never happen unless the numbers work. For all I know, it is not as tough as I think it is.
For my eighth wish, I think it is time to bring the DB2 BLU Accelerator columnar data store and database acceleration techniques to the DB2 for i relational database that is at the heart of the IBM i platform. DB2 BLU Accelerator, which made its debut back in April 2013, is used to speed up reporting and analytics jobs that rely on relational databases and uses data compression, in-memory, and other techniques in conjunction to accomplish this. If the Linux, Unix, and Windows variant of DB2 has the BLU Acceleration features, then the IBM i-DB2 combo should. It is that simple.
Finally, I want IBM to provide incentive pricing for IBM i software and related system tools when customers add new workloads to their systems, as it has been doing for mobile front-ends on its System z mainframes since last year. These discounts run up as high as 60 percent for software licenses on capacity dedicated to such new workloads. If IBM added LASE and WASE environments and allowed customers to bring Linux and Windows workloads onto IBM i, the customer should get a deal for moving stuff onto the box rather than off it.
Now, you have to do your part. Send me your IBM i wish lists for 2015 so we can get even more things we think are necessary to support the IBM i platform.