Can open source still be a 3000 stream?
November 23, 2005
The 3000 community's most seasoned vets have spread their wings well beyond MPE in recent years. These are developers, founders of software vendors, even the top lab experts still inside HP. Many have been scouting the heady waters of open software such as Linux distributions and open source tools such as the scripting language perl. Much of this is available on the HP 3000, but it is in need of attention.
Not long ago, a discussion among these vets raised the question, "How much trouble will the HP 3000 have following the open source stream?" The answers included observations about how much attention any kind of open source solution, even Linux, still demands from the brightest in your community.
Perl was the trigger for this discussion. David Greer, once a principal at Robelle until he sold off his share for a two-year journey by sail around the Mediterranean, posed the problem clearly. In a message he shared with us, he tried to set the stage for a study everyone in the 3000 community needs to undertake:
We are in an interesting time when I think the management of many of the common platforms (Windows, Linux, Sun, AIX, HP-UX, MPE) is getting much harder. To really run these well, you need to understand systems at a deep, deep level, just as we did in the 1980s with the HP 3000. But market dynamics suggest that no one is willing to fund this level of expertise. And no one is training people on a multi-platform basis, despite the fact that the Internet is forcing unheard-of interoperability."
Greer, who we profiled six years ago while he was president at Robelle, is working with tech startups today and experimenting with such open source staples as perl. He shared his own experiences with perl as a way of showing that open source will cost less — but it demands an investment of training and experience. Those are things the HP 3000 needed, too, according to another veteran of the platform.
Greer said:
I invested another three hours this afternoon to try and install the Perl modules that I was trying to install last week. Based on the excellent advice of using CPAN as a service and not just a Web site, I made a good go of it this afternoon. Hours later, I still couldn’t get everything installed. One tiny piece of the whole picture failed to install due to testing errors. The software engineer in me was thinking “good thing they had a test suite.”. The business manager in me was saying, “why can’t I just get my job done?”
To be fair, the company I’m working with, whose Web site is totally based around this technology, had warned me that the Perl Template library was known as a really big problem to install. In this company, I’m working with some of the best Perl programmers on the planet (e.g., Stas Beckman, Director of Development, has been the mod_perl implementer for the last three years and author of the O’Reilly mod_perl book). I can’t distract them from their primary job, but even someone like me with deep technical expertise can’t get to where I need to go (i.e., follow in the footsteps of these young, but experienced, developers).
To me, there is a deep underlying infrastructure problem which we are only on the edge of understanding. It is going to take some clever thinking to manage this infrastructure build-out going forward (HP 3000 or no HP 3000). No one has managed thousands and thousands of independent development teams so that the end result of their collective efforts can be truly integrated together in a reliable way.
Greer notes that the 3000 — in an era when the resources in HP's 3000 lab pros are now focused on other platforms as the company ramps down its MPE business — is going to need help from some source to keep open source software a reliable tool.
While I’m not focused on the future of the HP 3000 as a market, I can’t help but still think about it after spending more than 20 years on the platform. To me, this issue is the biggest short- to medium-term concern that I have with the viability of the HP 3000 platform: If it is to continue to survive, it must interoperate with the massive open software and Web services movement that is out there. It’s clear that the software piece is already getting a lot more difficult. I don’t know how this will play out, but my gut is saying that without solutions to this issue, HP 3000 customers will have to look elsewhere sooner rather than later.
Since this is a serious subject whether you are migrating or homesteading, we'll look closer at this tomorrow. But it seems that the community needs a lab effort to work on even the open source pieces of MPE/iX — Samba, perl, PHP, sendmail, PGP, Java and more. The alternative could be moving to Linux, built entirely of open source developers. Some 3000 veterans, however, report that Linux needs a more attention to work in place of a vendor-supplied environment.