Why Emulation Doesn't Compute For Us
January 26, 2009
[Ed. Note: We asked the 3000 community to tell us if a 3000 hardware emulator, powered by the last release of MPE/iX, would be of any use in the years to come. Many said yes, or perhaps. But one 3000 customer from a Canadian shipping brokerage makes reasoned points for why strapping MPE onto new, faster silicon, plus carrying MPE-based solutions along, is a poor fit. We'll have more emulator responses in our February print edition, and on the blog afterward.]
By James Byrne
Here at Harte Lyne we have two HP 3000 918LX systems, primary and a hot spare at our off-site location. I have read the articles and commentary of last week and, even allowing for my profound dissatisfaction with HP, my reaction to them is “more of the same old-same old... divert, deceive and delay.” We are not considering using an emulator for our HP 3000. This decision is based on three considerations:
One, as pointed out, there is no such emulator. It is more than seven years since the EOL announcement for the HP 3000. If an emulator was going to appear then one reasonably expects that one would be produced by now. Two, HP has demonstrated an intractable institutional resistance to admitting that the HP 3000 was a viable platform despite their own 2001 assessment to the contrary. This has had, and cannot but continue to have, a baleful influence on efforts at cooperation with HP by those producing and intending to use said (non-extant) emulators.
Three, emulation is not enough. The world has moved on considerably since 2001, while MPE/iX has not. Basic FTP and Telnet are inherently insecure and increasingly discarded methods of data transfer. SSL with SHA2 or SHA512 encryption is a de facto, and in many instances a de jure, requirement for business data communication between hosts and even for inter-process communication on unsealed servers. Compiler-driven languages are all but completely replaced in new business application development by interpretive, and processor intensive, virtual machines (IVM) such as Java, PHP, Python, Ruby and so forth.
An MPE/iX emulator, given the OS’s dated capabilities, would be a hard sell for most company’s IT departments, even if it and the license transfer were free. Having to pay for either, and no doubt facing considerable third party fees to transfer licenses like Cognos and such, makes this path a non-starter in all but what can only be a very few extreme cases.
Consider the device that I am using to compose this message, in a browser window. The host I am using has multiple CoreDuo2 64 bit processors with 4 MB of L2 cache and FSB. It has 16 GB of 64-bit memory, 1 TB of disk storage and a DVD RW-RAM multidrive. It is a generic Intel “whitebox” and its total cost was $800 CAD.
It is running a Linux OS, CentOS-5.2, which is FREE. On this device, which also acts as my desktop PC, I am developing the application software that will enable the last of our business systems to migrate off the HP 3000. This project uses PostgreSQL as the DBMS (FREE), NginX high performance httpd (FREE) as the service interface, together with the Ruby programming language (FREE) and the Ruby On Rails ORM framework (FREE) for the application software. We are also using Redmine project management software (FREE) and GiT distributed version control system (FREE) to control and document this project. When deployed it will use the Firefox browser with the xforms plugin as the client software (both FREE).
What would the equivalents of these products cost on an HP 3000? Tens of thousands, hundreds of thousand of dollars? For each separate host? Will any MPE/iX emulator be permitted by HP to run on an open source OS and commodity hardware? Or, will it be constrained to HP-UX on HP 9000s, with commensurate licensing fees for that OS and inflated prices for the hardware to run it?
How can an emulator of an operating system whose entire support structure is rooted in an outdated, and increasingly irrelevant, price per incremental feature marketing philosophy compete with the reality of commodity-priced hardware and open source free software? Whatever its value might have been five or six years ago, in my opinion desire for an MPE/iX emulator is principally driven by sentiment at this point. Sentiment is not a sound basis for making decisions.
I do not agree with HP’s decision to terminate the HP 3000. That said, without a revolutionary overhaul of HP’s marketing, product positioning and pricing with respect to this technology, then the outcome nonetheless would have been the same, the demise of the HP 3000, as will happen to HP-UX and the HP 9000 eventually.
Such a revolution had no chance of happening inside HP. For years, decades, prior to 2001, HP was advised by many, far more knowledgeable and influential than myself, that it was HP’s own internal policies that were strangling the HP 3000. As the song goes, “They would not listen / they’re not listening still / perhaps they never will.”
We no longer employ HP products, save for equipment that we already owned prior to November 2001. HP has taken its own path and we no longer travel in company with it. It is rather doubtful that they will ever notice.