NBAR and HTTP Data Conversations

I’m still working on the ONT test and doing labs, so I marked up a lab for me to work.  I’m using the same setup as I did last time.  The two routers are 3640s running 12.4(25b).

nbar-classmap1

Part of the lab was to identify HTTP traffic coming into F0/0 and mark it as CS3.  That’s pretty easy, right?  Of course, the lab I made up was a little more complicated, but the point comes clear with a simpler example.

class-map match-all HTTP
 match protocol http
!
policy-map PM-F0/0-IN
 class HTTP
  set dscp cs3
!
interface FastEthernet0/0
 service-policy input PM-F0/0-IN

I fired off a small script on TestHost1 to repeatedly do NMap scans on TCP/80 of TestHost2 to generate some traffic. 

root@bt:~#while ( true ) do nmap -sT -p 80 -v -n 172.16.2.101; done

I let that run for a while and checked out the service policy on F0/0; there were absolutely no matches on that class.

R1#sh policy-map int f0/0 input class HTTP
 FastEthernet0/0

  Service-policy input: PM-F0/0-IN

    Class-map: HTTP (match-all)
      0 packets, 0 bytes
      5 minute offered rate 0 bps, drop rate 0 bps
      Match: protocol http
      QoS Set
        dscp cs3
          Packets marked 0

I thought that was strange, so I kept the script running and captured the traffic coming out of S1/0.  Looking at the packets in Wireshark showed that none of the HTTP packets were showing up as being marked as CS3; they’re all set to the default DSCP value.

HTTP-nomarkNo matter how many NMap scans cycled through, the class never incremented a bit.  I let it run for two full minutes and generated a few hundred HTTP packets, but there was still nothing.

On a whim, I enabled NBAR protocol discovery on F0/0 to see if that would shed any light on my mess.  Guess what I found.  That’s right; there were no HTTP matches to NBAR, either.  That makes sense since the class-map I defined uses the NBAR protocol.  Alright, so it seems that it’s NBAR that doesn’t see the packets as HTTP, but why?

Looking through the packet captures again, I noticed that there was no real data in the streams.  I saw the 3-way TCP handshake (a.k.a., my signature wrestling move) and then a RST,ACK.  I only told NMap to check if the port was open, so I changed the while loop a bit and enabled version detection with the “-sV” flag.  This time when I can the script, NMap was actually getting the HTTP banner.  It was much traffic, but it was an actual HTTP conversation, so I checked NBAR again.  Success!  The same for the class, too. 

For craps and smiles, I created a class-map that matched SIP, added it to the same policy-map, and set NMap after TCP/5060 without version detection.  Without having a real SIP conversation, the class counters incremented as long as I was sending packets.  That would seem unexpected until you realize that NBAR has some advanced knowledge of HTTP; you can actually match on URLs, hostnames, and MIME-types.  I guess that means it also know when a real HTTP conversation is taking place.

To finish out the testing, I added an ACL to the router that matched any-any-eq-80.  I made the class-map into a match-any and added the ACL.  Since the ACL just matches the destination port and doesn’t care what the content is, every packet sent matched the class as expected.  I remember reading several places and seeing a couple videos that said that you can use NBAR matching and ACLs interchangeably.  That may not really be true when it comes to HTTP.

Send any Cisco learning credits questions my way.

jac