On Wed, 08 Dec 2004 14:16:52 +1300, Jason Haar <[EMAIL PROTECTED]> wrote:
> 
> I just ran it here - nessus client logon generated 5.3Mbytes of traffic
> over a WAN at an average of 0.24Mbs. I also have iperf installed on the
> Nessus server - it rated 1.2Mbs to the same workstation with Nessus
> client... That's 5 times faster...
> 
> It seems nessusd is the bottleneck there - not the network? I don't have
> majorly fast processes (P-III), so I'm sure a bigger CPU would help, but
> I wonder why it takes nessusd so long to push that data out to the client?
> 

The box I picked wasn't the best example - I hadn't used that
particular one in quite a while, and it was still running 2.0.x. 
Upgraded to 2.2.0, and the plugins are now ~5.3 MB.  It can still fill
a 1 Mb link though.  But we're talking a dual P3 1 GHz with SCSI 10K
RPM drives.

There's a huge difference between iperf (just pumping out as much data
as you can) and loading ~5800 Nessus plugins to the client.  For the
latter, nessusd has to open ~5800 nasl files (if I understand how it's
working), which can take a while, especially if you're using IDE
drives.  If I had to guess I'd say disk I/O is your limiting factor. 
A P4 1.6 GHz with an IDE drive wasn't a whole lot faster over gigabit
LAN than the box with SCSI drives over 1 Mb VPN.  That's what I'm
basing my guess on.  Both are FreeBSD 5.3 boxes.

-Chris
_______________________________________________
Nessus mailing list
[EMAIL PROTECTED]
http://mail.nessus.org/mailman/listinfo/nessus

Reply via email to