How Qwest Runs Things

Brew Bird writes "Qwest explains how they handle the various issues that crop up being a large ISP/Backbone Provider. They've got the presentations setup in a nice little website." It's very *BSD focused, since I believe that's mostly what Qwest runs but the presentations are interesting in the scaling issue - what do you do with that much data and that many machines?

Comments

second-hand info

[The_Messenger]

My boss used to work in IT at several of Qwest's East Coast shops, and according to him the vast majority of Qwest's business backend runs the standard 'Solaris/Oracle on Sun' setup. Any BSD that Qwest uses is likely confined to the actual routers and switches,not only because BSD doesn't run on the type of hardware needed for their amount of traffic, but BSD doesn't even the right software.

Also, I used to be a BSD nut, so while I think it's cool to hear about its use at such large corporations, understand that 80% or more of it will be BSDi. In several areas FreeBSD technically out-performs BSDi's server products, but BSD has both a corporate reputation and a full-time, in-house support staff. Stories about BSD's success in the enterprise are not "news" and should not be considered good for Free Software.

Last year's BSDi/Walnut Creek merger affected FreeBSD immediately (as I'm sure other BSD users noticed between 4.1 and 4.2), but I'm not yet sure if BSDi's Internet Server OS has been affected.

One thing that many inexperienced Slashdotter's don't seem to know is that while BSD and GNU/Linux are great (even superior, in many cases) for small-medium Intel boxes, there is a point where you really do need the high-end hardware and industrial-strength UNIX that only IBM, Sun, HP, and others currently provide. And it's not just a matter of corporate bullshit... for instance, ask anyone who has ported "real" enterprise software from UNIX to GNU/Linux about the experience and I guarantee that the first they'll do is start bitching about GNU/Linux's lack of a real threading model and other deficiences.

I hope that IBM's support of Linux continues, because that is truly one company whose knowledge and experience can help Linux overcome these issues.

This is getting way offtopic, but I'm going to mention that I think a Linux kernel fork is inevitable within the next five years. One group, led by De Icazza and others, will concentrate on bringing Linux to the level of the Windows desktop. The other group, led by IBM, TurboLinux and other corporate interests will concentrate on bringing Linux to the level of UNIX as an enterprise OS running on non-trivial (ie non-IA) hardware.

And ten years from now, after the Linux communities have destroyed each other, BSD will still be running network hardware in machine rooms around the world. :-)

All generalizations are false.