Re: Sockets: synchronous I/O

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Tue, Aug 02, 2005 at 12:33:33PM -0400, Pradeep Padala wrote:
> >Tux was a nice experiment, until somebody showed that you can get the
> >same performance from userspace by using sendfile(). The performance
> >argument no longer holds.
> 
> Agreed, but I believe there are still some gray areas. For example, 
> using multiple streams for transfering a file. Yes, this can be done in 
> the user mode, probably as effectively, but doing it in the kernel is 
> still an interesting thing to do.

If you mean "interesting" as in "interesting ways to crash your kernel
by minor programming errors", I agree.

> Another scenario is to perform data 
> transfers as close to wire speeds as possible.

Try vsftpd. It does wire speed transfers from userspace. It's the FTP
daemon that powers a couple of the largest FTP sites in the world, like
ftp.kernel.org, ftp.redhat.com, etc. To quote Alan Cox: "finally we
have a scalable ftpd for Linux". See http://vsftpd.beasts.org/ .

> I work in high 
> performance computing environments, and somethings can be done only by 
> tweaking the kernel.

There is a difference between tweaking and adding new functionality.


Erik

-- 
Erik Mouw
J.A.K.Mouw@xxxxxxxxxxxxxx  mouw@xxxxxxxxxxxx

Attachment: signature.asc
Description: Digital signature


[Index of Archives]     [Newbies FAQ]     [Linux Kernel Mentors]     [Linux Kernel Development]     [IETF Annouce]     [Git]     [Networking]     [Security]     [Bugtraq]     [Yosemite]     [MIPS Linux]     [ARM Linux]     [Linux RAID]     [Linux SCSI]     [Linux ACPI]
  Powered by Linux