lkml.org 
[lkml]   [2000]   [Jul]   [14]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Multithreaded TCP/IP stack


"Jeff V. Merkey" wrote:
>
> I think the source of this was probably the Mindcraft tests and
> Microsoft's website. If you visit the MS website, there's a bunch of
> inaccurate statements about why Linux with Apache was slower than NT
> with ISS due to the fact that the IP stack was not multi-threaded. This
> was a true statement about the Mindcraft NetWare vs. NT performance
> shootout since NetWare's lack of a multi-threaded IP stacks was why
> NetWare was slower with IP than IPX (since the IPX stack in NetWare is
> multi-threaded but the IP stack was not). For some reason, Bruce Weiner
> applied the same arguments to the Linux vs. NT shootout last year and MS
> picked up on his statements. The message got further confused when PC
> Week published information (along with Microsoft) that Linux was not as
> good as NT because the IP stack in Linux was not multi-threaded.
>
> What's in Linux is multi-threaded to some extent, but the real source of
> the performance difference between the two was due more to the fact the
> MS has implemented some special Ring0/Ring3 traps just for Bruce's
> benchmarks between IIS and their kernel to make their IIS server beat
> Linux (much like they did with client side caching and oplocks they used
> for the NetWare vs. NT shootouts) whereas Linux runs Apache as a true
> user mode app.


BTW. These traps allow the kernel in NT to initiate a ring transition
"upward" into specific Ring 3 applications to allow the kernel to pass
callbacks directly into the IIS server code rather than wait for the
next "poll" event on the IP sockets interface in NT. It works by
elminating latency for ACK packets .... One of my good friends up North
was bragging to me about how they stomped Linux's ass by putting in this
optimization and how their Ring3 mechanism with IIS was better than
Apache Linux due to this capability .....
:-)

Jeff

>
> This belief is mostly fud, as observed by someone else on this list.
>
> :-)
>
> Jeff
>
> Chris Wedgwood wrote:
> >
> > On Thu, Jul 13, 2000 at 01:47:08PM -0400, Richard B. Johnson wrote:
> >
> > On Thu, 13 Jul 2000, Stephen Torri wrote:
> >
> > > A friend of mine is critical of linux for not having a
> > > multithreaded TCP/IP stack. Is that true? If it is not then
> > > what kernel version introduced it. If it is then when should we
> > > expect to see it?
> > >
> > > Stephen
> > >
> >
> > This is cute. Could you ask your friend what a multi-threaded
> > TCP/IP stack is?
> >
> > His friend probably refers to the locking used by the stack and how
> > much concurrency you'll get with multiple processors.
> >
> > With 2.2.x most of the stack is under a single lock, with 2.4.x
> > locking will be such that multiple processors can be in the stack at
> > the same time (the performance should be much greater then under high
> > load on big SMP boxes).
> >
> > --cw
> >
> > -
> > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> > the body of a message to majordomo@vger.rutgers.edu
> > Please read the FAQ at http://www.tux.org/lkml/

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:57    [from the cache]
©2003-2014 Jasper Spaans. Advertise on this site