Snort mailing list archives

Re: Sizing of a box requiring 2x10Gbps


From: Russ Combs <rcombs () sourcefire com>
Date: Thu, 8 Jul 2010 15:35:53 -0400

Good points Will.  Some clarifications / comments below.

Russ

On Wed, Jul 7, 2010 at 7:06 PM, Will Metcalf <william.metcalf () gmail com>wrote:

For the most part I agree with what you guy's are saying although
there are some things that we have learned from working on the
"IDS-Who-Must-Not-Be-Named" that you could do to reduce packet loss of
stock snort that I'm actually really surprised you haven't done
to-date.

1. Support reading multiple packets per pcap_dispatch() call.


Snort has been doing this for a while (maybe since 2.7.0?).  Probably not
when you last looked at it though.


2. Along the same lines since libpcap-1.0 if the functionality is
available in the linux kernel, libpcap will use a mmap'd ring
buffer(essentially phil woods patch was integrated).  The size of this
buffer can be set via pcap_set_buffer_size() but you don't allow this
as an option to your users.


Snort has been bogged down with an old pcap for quite a while.  However, the
memory mapped pcap has been an option.  Snort 2.9.0 does in fact jump up to
libpcap 1.1.1 and provide a way to configure via pcap_set_buffer_size().

2.9.0 goes even further here.  It introduces a DAQ (for data acquisition),
basically an API and a suite of pluggable modules for packet acquisition and
injection.  In addition to using pcap, an afpacket module provides inline
memory mapped performance and there are a number of other modules as well,
including ipfw, ipq, and nfq.


3. Since I have a feeling you will never open source your in-kernel
load-balancing juju that you reference here

http://vrt-sourcefire.blogspot.com/2010/06/single-threaded-data-processing.html
(I'm
going to buy you a nehalem chip btw), why don't you guy's add support
for PF_RING cluster-sockets?  This way you can be happy with your
FUDless single thread while still allowing your users to load balance
based on flow by simply firing up multiple disparate snort processes
with the same PF_RING cluster id.  The time that it would take to
build this new packet acquisition method would be minimal.


Creating a PF_RING DAQ module is worth investigating.


Just my 2 cents..

Regards,

Will

On Wed, Jul 7, 2010 at 6:18 AM, Joel Esler <jesler () sourcefire com> wrote:
It would be very difficult to achieve those kinds of speeds without a
commercial Snort appliance like Sourcefire.

Sorry for the plug.

--
Sent from my iPad

On Jul 7, 2010, at 4:28 AM, "Sven Juergensen (KielNET)" <
s.juergensen () kielnet de> wrote:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi list,

I'm playing with the thought of implementing an
IDS for our network. Now, for the box handling
this, a bit of advice would be appreciated. It
needs 2 10GE interfaces and would have to soak
up a throughput of about 4GBps tops. The amount
of accumulated data should last about a week.

Does anyone know the rough specs for a box to
deal with this?

Thanks in advance and regards,

Mit freundlichen Gruessen,

      i. A. Sven Juergensen

- --
Fachbereich
Netze und Rechenzentren

KielNET GmbH
Gesellschaft fuer Kommunikation
Preusserstr. 1-9, 24105 Kiel

Telefon : 0431 2219-053
Mobil   : 0170 403 5600
Telefax : 0431 2219-005
E-Mail  : s.juergensen () kielnet de
Internet: http://www.kielnet.de

Geschaeftsfuehrer Eberhard Schmidt
HRB 4499 (Amtsgericht Kiel)

PGP details at
http://pgp.kielnet.de/sjuergensen/

-----BEGIN PGP SIGNATURE-----
Version: GnuPG/MacGPG2 v2.0.14 (Darwin)

iEYEARECAAYFAkw0Or0ACgkQnEU7erAt4TLYvQCgro8f56KLyt6QH4gSql4GO8CS
c+8AoPJfbeK3Ft+jgqmv3gFih7K41tkw
=8R2s
-----END PGP SIGNATURE-----


------------------------------------------------------------------------------
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
_______________________________________________
Snort-users mailing list
Snort-users () lists sourceforge net
Go to this URL to change user options or unsubscribe:
https://lists.sourceforge.net/lists/listinfo/snort-users
Snort-users list archive:
http://www.geocrawler.com/redir-sf.php3?list=snort-users


------------------------------------------------------------------------------
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
_______________________________________________
Snort-users mailing list
Snort-users () lists sourceforge net
Go to this URL to change user options or unsubscribe:
https://lists.sourceforge.net/lists/listinfo/snort-users
Snort-users list archive:
http://www.geocrawler.com/redir-sf.php3?list=snort-users



------------------------------------------------------------------------------
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
_______________________________________________
Snort-users mailing list
Snort-users () lists sourceforge net
Go to this URL to change user options or unsubscribe:
https://lists.sourceforge.net/lists/listinfo/snort-users
Snort-users list archive:
http://www.geocrawler.com/redir-sf.php3?list=snort-users

------------------------------------------------------------------------------
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
_______________________________________________
Snort-users mailing list
Snort-users () lists sourceforge net
Go to this URL to change user options or unsubscribe:
https://lists.sourceforge.net/lists/listinfo/snort-users
Snort-users list archive:
http://www.geocrawler.com/redir-sf.php3?list=snort-users

Current thread: