You are here

The Monkeys Chased Away Puffy, or, Replacing OpenBSD with NetBSD at the Point of Sale

Submitted by Marc Balmer on Mon, 12/21/2009 - 12:13

About one year ago, I wrote about how we used OpenBSD at the point of sale at the Basel Zoo. Well, the good news first: We replaced OpenBSD by NetBSD for our POS applications, both in development, but also for deployment. That might sound like a radical step, but in fact it is not. It is just the consequence from OpenBSD not being ready for prime-time in an enterprise world. The lack of proper SMP support, for one, is a major problem with todays computers being mostly multi-core. And for CPU demanding applications like databases that process POS data we just can not afford having an idle CPU for almost all the time. Another problem are file systems. No support for modern filesystems, not even FFS2. And the NFS performance is not really what we need. The POS systems are started every morning on a schedule that takes into account visitor frequencies. A central system boots the terminals using Wake-on-LAN. NetBSD (as well as FreeBSD, btw) has the wake(8) command to send Wake-on-LAN packets over the network*. The wake(8) command, which I wrote early this year, helps us to save energy and is just generally very practical to power-up remote machines, either automated, or manually, e.g. when you have no physical access to a machine. There were other, smaller issues, that don't really matter to much here (installer issues, lack of i18n in the base OS, etc). In retrospective, it was quite a bit of work to make the switch; but I am happy we did it. Performance got a lot better. Installation got easier. The firewalls the protecting the POS systems are still OpenBSD, since the machines are single-core only and pf is just the best firewall. I hope one day someone will update pf in NetBSD and FreeBSD to a more recent version. * wake(8) was in OpenBSD as well, but unfortunately only for a short period of time. As you all know by now, the OpenBSD project leader forced me to remove it under the stupid pretext that "the bin directories are full."... This forced removal was one of the reasons for me to leave the project, since it was not the first time I was forced to backout good stuff for no reason. When a group has a single person that has the power to force developers to back out stuff that was liked and OK'ed by many, can that really be called a group? I don't think so and I definitely do not want to be part of such a constellation.