## On 2002-04-27 14:40 +0200 Mikael Olsson typed:

MO>
MO> Rafi Sadowsky wrote:
MO> >
MO> > [on syslog performance and delayed writes]
MO> >  Wouldn't you agree that benchmarking the effect on Disk I/O is at
MO> > least as relevant as the effect on the CPU ?
MO>
MO> If this box did a lot more than receive syslogs -- yes, it would
MO> be relevant. However, an overloaded disk i/o bus results in
MO> the the disk i/o calls blocking, which results in higher CPU,
MO> so in the simple case

 I assume that indicates one EIDE disk ?

MO> , low CPU == no real problem with disk i/o.

[This is not an OS advocacy invitation -
 though feel welcome to correct me if I'm wrong]

 You didn't answer whether it's Linux or not - for Linux & Solaris indeed
You're(probably?) right but for(example) FreeBSD the CPU usage doesn't
(seem to) climb in direct correlation with disk I/O blocking


MO>
MO> And 1GB per day isn't a problem for _any_ modern disk.

 Oh come on - I'm sure you realize that the disk seek time is the limiting
factor ant not the peak/burst transfer speed
(assume syslogd is logging to more than one file)

MO>
MO> Heck, even floppy can save about 1MB a minute, and there are
MO> 1440 minutes per day, so even assuming a disk as slow as
MO> a floppy, you should still be able to receive more than 100
MO> syslog entries per second. (Although in _that_ case, you _would_
MO> see high CPU loads ;))
MO>
MO>
MO>

_______________________________________________
Firewalls mailing list
[EMAIL PROTECTED]
For Account Management (unsubscribe, get/change password, etc) Please go to:
http://lists.gnac.net/mailman/listinfo/firewalls

Reply via email to