IntlHarvester's Journal: *BSD faces a very bleak future.

*BSD faces a very bleak future. I've seen the same boring cut-n-paste "BSD is dying!" trolls for years now too, so don't dismiss what I have to say as another one of those. I researched many compartive points about all the various flavours of *BSD after my comptroller asked me to deploy an OpenBSD firewall.

Granted 4.2BSD was a very fine OS, but that was in 1983. 4.4BSD, and its brother 4.4BSBD-Lite, were abymsmal performers at best during their heydey in 1993-4. Both Solaris and HP-UX had networking stacks that supported "long fat pipes," multicasting, and TCP header header prediction years before 4.4BSD did.

I don't know why 4.4BSD-Lite became so popular. Perhaps because it was released as OpenSource in 1994? But even then there were much better TCP/IP stacks and VM schemes in use (Solaris, AIX) so availability of source code was an insignificant win at best. All OpenSource does is allow poor quality code to be re-circulated and reused again and again in new systems, while high quality and RFC compliant code is relagated to the pay environment.

Regardless, the codebase of 4.4BSD-Lite became the stepping stone for all the *BSDs that are still around now. The main three *BSDs (FreeBSD, OpenBSD, NetBSD) all use at least 85% of 4.4BSD-Lite's source code, with the rest being mostly new userland code, TCP/IP updates, and multiprocessor support.

The commerical offering, BSDI, is even more appaling - a source code diff shows roughly 94% code reuse. Paying for an archaic and outdated OS...that would explain why BSDI has less than 2% of the server market.

FreeBSD has very close ties with BSDI. I'm not one to preach doom by association, but I'm afraid FreeBSD has doomed itself by the move. If that isn't enough, FreeBSD's C2 security certification is horrible. Even NT can do better than it!

FreeBSD has a reputation of being the "fastest" BSD on x86 hardware. Actual memory bandwidth performance is a fraction of all of Sun's offerings, and the multiprocessor support is a joke since it has a poorly implemented semaphore locking mechanism. I hear a total re-write is planned, and perhaps even a security audit too, so /maybe/ by 2005 FreeBSD will be a contender in the low-end server market.

NetBSD, I'm afraid, is dead before it got off the ground. The goal of running on as many platforms at once is a noble and idealistic one, but in the real world its useless. At best NetBSD is a mediocre hobbyist OS that runs on outdated computers. A match made in hell it would seem, since ancient source code has been hacked to run on ancient computer. Its ports to systems such as the Dreamcast are total folly, offering no more real world use than GUI systems on headless servers. And I think the installed user base of less than 10,000 speaks for itself.

I was hopeful OpenBSD would be better as its reputation for security is interesting. Sadly, its another strikeout. OpenBSD's filesystem is extremely slow, and hardware support is nearly nonexistant. There are also numerous political issues surrouding its development team that are eating away the last bit of hope. Perhaps the reason it is secure is because no one bothers to hack it since the "prize" is mostly worthless.

*BSD users too are dooming thier own OS. As a group, they are a very vocal and rowdy bunch. No real help is given to new users and such an elitest attitude is suicide.

I chose to not deploy an OpenBSD based upon these reasons. It is my humble opinion that either NT or Solaris be used for any significant work, and *BSDs be left to the hobbyists.