On 2011-12-15, Kostas Zorbadelos <kzo...@otenet.gr> wrote: > Greetings to all, > > we are running a project to anycast our DNS resolver infrastructure. The > case is a big commercial country-wide IP network. The company uses Linux > extensively in the infrastructure but no BSDs. > > I keep an eye on OpenBSD developments (mostly high level) and use the > system personally, but I have no personal experience in larger setups and > production services. I find the project a good match for OpenBSD, > because of the system's strong networking features and routing > support. I will definitely include OpenBSD in our tests and hopefully > make a case for it, to introduce it in our infrastructure. > > The main contenders as you realise are Linux-based setups with either > Quagga or BIRD. As for DNS software we will stick with BIND for now and > perhaps consider UNBOUND in the future (when the future involves > DNSSEC). From what I have seen so far in various sources, people mention > Quagga's scalability problems and maybe old architecture while good > words are said about BIRD. We are after a solid OSPF implementation both > v2 and v3 (IPv6). I have seen OpenBSD's routing software architecture > and I like it a lot and I also have a high regard for the system's > quality. > > Of course personal taste is not enough as you understand to support a > case of introduction of a new platform in a production, commercial > environment with A LOT of contraints mostly non-technical. The questions > therefore are: > > - has anyone done anything similar using OpenBSD that would like to > share? > > - how would you compare with facts and not flamewars OpenOSPFd against > Quagga or BIRD implementations?
Quagga doesn't seem to care much about OpenBSD, the current version doesn't even build here. (I did port the last round of ospf crash fixes to the previous version which does build, these are in the ports tree). Development is very fragmented, a lot of tweaks exist in 3rd party repos but there seems to be no central group trying to hold them together (at one point it looked like the google fork might do this but it appears to have stagnated). OpenBSD is not a primary target for BIRD, it runs here but misses some things. I'd probably favour it over Quagga if for some reason you couldn't use the native tools. I certainly wouldn't have a problem running something like this using ospfd. I haven't used ospf6d at all yet myself so couldn't comment on it. > - what is your opinion about using a latest version of BIND from ISC > instead of the BIND distribution coming with OpenBSD? I haven't used either much recently, so I can't really say. (I'm running Unbound myself). > - is there any option of commercial support? There are some on www.openbsd.org/support.html who can help, but if you need something particular you might do better to outline what you need on misc@ and ask for people to reply off-list ... > - would you consider Java support on OpenBSD "production quality"? Seems > irrelevant but we might utilize some Java tools for > measurement/statistics I've only used it for non-critical things myself but it seems to work fairly well. > Thanks for the very good and hard work on the system. > I would be interested to hear any thoughts even off-list. > > Regards, > > Kostas
