[Snort-users] Snort Performance
erek at ...577...
Tue Sep 10 09:41:02 EDT 2002
On Tue, 10 Sep 2002 jsp1999 at ...348... wrote:
> Snort is a great tool that offers convenient ways to customize the network
> traffic that should be monitored.
Errr... Snort is an intrusions detection program. Network monitoring needs
to be left to the likes of ntop and iptraf.
> Unfortunately we found out that there is a big problem if nearly all the
> available rules are used during operation.
Yes, that's right. Why do you think we say "tune your ruleset for your
> Snort does not look at all the packets, it often simply skips packets. On
> a highly loaded network this gets worse - more and more packets are simply
> not analyzed.
> Isn't this very dangerous, because many exploits require only a few
> packets to perform an exploit and to compromise machines?
Again, that's why you tune your rules. :) Check the archives, that's been
said from day one.
> When we had an in depth look at the source code of snort, we saw that
> there are the RTN and OTN structures for storing the individual rules
> which have to be iterated through every time a new packet is matched.
> We consider this not to be optimal, because of the many ANY parts in the
> source and destination IPs. Has anybody thought about optimizing this
> basic data structure? Will this be improved in Snort 2.0 (we found some
> PPT presentations on the web)? Are there any chances for improving the
> ratio of investigated packets / actual packets on the network?
Why is it not optimal? Care to elaborate? As for the 'any parts', well...
We often tell folks to put the more specific rules at the top of the config to
improve performance. Please check the archives on this--It's there. :)
Snort 2.0 will be different. Many things will have changed. But, we're not
there yet. We've got to get 1.9 out of the door before 2.0 can be worried
More information about the Snort-users