Originally posted by profoundWHALE
View Post
Announcement
Collapse
No announcement yet.
Facebook Is Hiring To Make Linux Networking Better Than FreeBSD
Collapse
X
-
Originally posted by david_lynch View PostThe statements that "Linux is the fastest" and "We want to make it faster" are not mutually exclusiveLast edited by jimbohale; 21 August 2014, 06:46 PM.
Comment
-
Originally posted by jimbohale View PostIf it's fast and the CPU load is still high then there's a problem. That's where I suspect they are targeting. For me, my NAS bottlenecks on Linux and not FreeBSD due to CPU usage because it *is* low-spec'd, but it's not like it's that low. It's just an atom processor, and I'm quite certain atom's are useful in servers due to their low power consumption. I get straight gigabit transfer speeds with FreeBSD with relatively low CPU load, 50 MB/s at the most (on any service) on Linux and CPU load at 100%.
Comment
-
Originally posted by david_lynch View PostTo be sure, there are further opportunities for optimization in the Linux kernel. BSD has been doing things right for many years.
Comment
-
Originally posted by Pawlerson View PostKeep dreaming. Linux beat it years ago. That's why companies migrate their servers to Linux and that's why Linux server market share is MUCH higher than BSD:
http://en.wikipedia.org/wiki/Usage_s...n_the_Internet
Linux has basically a bad network stack.
Linux networking was significantly improved after famous Mindcraft fiasco . The latter took place in the early 1999 and was related tot he results of Microsoft sponsored (and Mindcraft executed) tests that showed that despite Raymondism claims, Linux 2.2 had problems in the application area were it is most widely used -- as a web server. Windows NT simply wiped the floor with Linux. Here are some important results from the test (which was devastating to the pride of linuxoids, and served as an important stimulus for the improvement of the subsystem):
With 4 CPUs and 1 Gig of RAM, NT & IIS achieved 4,166 http requests per second.
With 4 CPUs and 1 Gig of RAM, Linux & Apache achieved 1,842 http requests per second.
With 1 CPU and 256 MB RAM, NT & IIS achieved 1,863 http requests per second.
With 1 CPU and 256 MB RAM, Linux & Apache achieved 1,314 http requests per second.
this site shows the issues.
Kernel issue #1: TCP bug
Gets hired to improve the Linux network stack, ends up migrating all servers from Linux to others.
Comment
-
Originally posted by nasyt View Post
Since 2000, the Linux kernel support for SMP has gotten better. The TCP stack has been improved. The scheduler has been improved. The ethernet card drivers have been improved. This is not the 2.4 kernel any more.
Apache has also improved - but if you're going to serve static content, today it might make more sense to benchmark against nginx.
I assume this was just the tip of the iceberg
Gets hired to improve the Linux network stack, ends up migrating all servers from Linux to others.
Comment
-
(Follow up)
To be fair, I've read interviews with Windows developer Mark Russinovich in which, in the late 1990s, he criticized the Linux kernel for lacking some forms of asynchronous IO, for only offering 'select' for waiting for events, for only having partial support for reentrant calls, and at the time the NT implementation of sendfile worked better with the filesystem cache than the Linux implementation of sendfile. (See: http://windowsitpro.com/systems-mana...se-01-apr-1999 )
I strongly suspect that at that time, there were a lot of Linux zealots who ignored his points and ignored other people who made similar points. But as far as I know, the current Linux kernel has addressed every one of those issues.
Comment
-
Probably because in other respects, Linux fits the needs of Facebook better. Not being Facebook, I cannot say why. Linux has wider hardware support, but that probably is a non-issue for a company that buys servers by the boatload, and therefore could specify devices that work well with FreeBSD. Perhaps their software has developed dependencies on the details of Linux.
Comment
Comment