What to Expect in Performance This Year
January 2, 2015
Legacy systems like the HP 3000 remain entrenched around the world. The reason? Their durability and their standing in the company's business legacy. What's a business legacy, you ask? It's the history of what kinds of servers and programs get used to process business. All MPE/iX applications are business legacies by now. They're more than a decade old. They run, and their performance is adequate. There seems like there's little to be done about making them faster.
But employing an emulator to replace the Hewlett-Packard models of 3000s can change that. The promise is more performance from more modern Intel-based hardware. There are limits, however. Here in 2015, the performance gain is limited by the size of 3000 that's running this week, the first of the new year. This week we read about "orders of magnitude" performance gains, but that's usually a number only applicable to a first order -- times 10. And even that might be a few years away for 3000 managers.
Given enough time, everyone who uses a 3000 emulator will outstrip the raw processing power of the HP-brand iron. Those HP boxes will never get faster, unless you can top them up on memory. In contrast, the Stromasys emulator will get more efficient; 2015 sees a newer, faster version now available. And Intel-based iron will grow stronger, too, at its top-end. The phrase "top-end" matters a great deal. If you're using top-end HP hardware, it might be too soon to look for a significant performance boost from virtualization.
Top-end means the fastest N-Class servers. Those will need to be replaced by top-end Intel hardware: servers with many available CPU cores, and many CPUs. Faster might not be a goal, however, for 2015. As-fast might be enough, to enable a manager can leave behind the aging HP iron.
It's easy to misunderstand. At a website called The VAR Guy, written by former InfoWorld editor in chief Michael Vizard, Stromasys' potential got noticed. "After all," he said, "the latest generation of Intel processors provide orders of magnitude more performance than VAX, Alpha, HP 3000 or Sparc systems that can be more than a decade old." Um, sometimes. But when you're working at the top-end of the old hardware, orders of magnitude is a far-off, wishful goal. If your HP 3000 has a tiny 3000 Performance Unit rating of 2.7, for example, then the first order of magnitude would be 27. The next order is 270, and so on. Several orders may be possible — at the lower levels of 3000 performance.
Simply beating the existing performance is still a valid desire, though. Matching what you're using — so you can leave old hardware behind — is a bona fide need in the 3000 market.
But The VAR Guy does more than overlook the real-world limits on the current virtualization product. It seems that virtualization is somehow a wedge into bigger replacement plans.
For solution providers, the ability to move those legacy applications to x86 servers should create an opportunity to have discussion not just about saving a few dollars, but more importantly, how that money might then be reallocated somewhere else in 2015.
As we come to the close of 2014, reducing legacy infrastructure costs is almost always top of mind for the internal IT department. Unless they can achieve that goal they typically don’t have enough funds available to allocate to new IT projects.
We're not sure what The Guy means by solution providers. Stromasys is a solution provider, but it doesn't have any interest in discussing the reallocation of money to other operations in 2015. The Charon solution is a replacement with a future that emulates technology driving the business. There will be enough new spending in a virtualization plan anyway, buying new Intel iron that's fast enough to match the old HP iron's performance.
What's certain is that "solution providers" doesn't mean the vendor of the legacy system. Not for the HP 3000. A few years ago, yes, HP was announced as a Global Partner of Stromasys. But we don't know of any stories where HP's introduced Charon to a 3000 site. As a vendor, HP's not going to help a 3000 site much if a customer installs Charon. The extended life of any MPE applications might give a customer more time to migrate.
Other parts of the solution lay in the apps. Some application vendors have abandoned their 3000 apps. Those who have left not solution providers, either.
In short, any solution provider who offers an emulator to reduce legacy infrastructure costs won't have ideas for how to better spend money being saved in dumping old hardware. The old hardware is paid-for. There's not an immediate savings in this equation, unless you can reduce 3000 hardware support spending, by a lot. You only get funds to allocate to new IT projects by cutting costs that don't require any investment to do those cuts. Reducing staff comes to mind. People want to cut out old hardware — not the old hands who know how to manage the hardware's OS.