Solved: GRE Encapsulation Question

lamav · ‎09-12-2008

I'd like a little clarification regarding GRE encapsulation.

From the way I have always understood it, an IP packet is encapsulated into/by a GRE packet that has a 20 byte header. But nowhere in that header exists a source IP or destination IP address field (The tunnel source and destination addresses would be the ones of interest to GRE). So, how does GRE forward the packet to the tunnel destination?

Moreover, it seems that, logically speaking, GRE sits between IP (network layer) and the data link layer. Is that correct?

I have yet to find any web link, including Cisco's, that really explains the nuts and bolts -- the mechanics -- of GRE to the degree I am describing here. Does anyone have one?

Thanks, folks

Victor

Istvan_Rabai · ‎09-13-2008

Hi Victor!

It's good to hear from you, too.

For the last several months I've been having a lot of other tasks to do and I have little time to participate in the forum.

However, when I find some time, I always return to this site because it's an excellent place to find challenging problems.

I found a good stuff for you:

http://www.ciscopress.com/content/images/9781587201509/samplechapter/158720150X_CH14.pdf

Cheers:

Istvan

View solution in original post

Richard Burts · ‎09-12-2008

Victor

My understanding of GRE appears to be very different from yours. Yes GRE involves a 20 byte header - and that is an additional IP header (what is the length of an IP header?). In the GRE header the source address is the address specified in tunnel source and the destination address is the address specified in tunnel destination.

If you have not done so before, I suggest that you try a packet capture (wireshark or whatever you prefer) on a GRE packet.

HTH

Rick

HTH

Rick

lamav · ‎09-12-2008

"In the GRE header the source address is the address specified in tunnel source and the destination address is the address specified in tunnel destination."

Rick, that's just it; there are no address fields in the GRE header, yet the GRE packet encapsulates the IP datagram and forwards it.

RFC: http://www.ietf.org/rfc/rfc1701.txt

What I think MAY be the case is that the IP datagram gets encapsulated by GRE, BUT the IP header remains "outside." So, what you have is the IP header, then the GRE header, and then the IP datagram inside the data portion of the GRE packet. Then ALL of that gets encapsulated by a L2 frame.

[EDIT] Better stated, I think GRE creates a NEW IP header with the tunnel souce and destination addresses, and THAT header's info is used to route to the tunnel endpoint. [EDIT]

This is what I think MAY be the case, although I cant seem to find any really good documentation on the nitty gritty mechanics of it.

What do you think?

Victor

Istvan_Rabai · ‎09-12-2008

Hi Victor,

This is the point:

"GRE creates a NEW IP header with the tunnel souce and destination addresses, and THAT header's info is used to route to the tunnel endpoint."

GRE header minimum length is 24 bytes, not 20 bytes: after the NEW IP header it contains a 2 byte flags field and a 2 byte protocol field that tells the next header information.

The protocol field's value is dependent on what type of packet is encapsulated. If IPv4 packet is encapsulated, then the protocol field contains 0x0800.

After this protocol field follows the original IP header and the rest of the original IP packet.

This is the link on cisco.com about GRE:

http://www.cisco.com/en/US/tech/tk827/tk369/tk287/tsd_technology_support_sub-protocol_home.html

Cheers:

Istvan

lamav · ‎09-13-2008

Hey, Istvan! Long time, man....where have you been?

OK, I'm glad that someone has validated my suspicion. Thanks for your answer.

I did go to that link you gave me before and it has about 30 links on it. They are all about configuring GRE. Which is the one that discusses the theory and the detailed mechanics of how it works?

Thanks

Victor

Istvan_Rabai · ‎09-13-2008

Hi Victor!

It's good to hear from you, too.

For the last several months I've been having a lot of other tasks to do and I have little time to participate in the forum.

However, when I find some time, I always return to this site because it's an excellent place to find challenging problems.

I found a good stuff for you:

http://www.ciscopress.com/content/images/9781587201509/samplechapter/158720150X_CH14.pdf

Cheers:

Istvan

lamav · ‎09-13-2008

Istvan:

Well, I'm glad then that you decided to come back to the forum exactly when I asked this question! LOL

That's an excellent link, by the way.

Thank you!

Victor

Istvan_Rabai · ‎09-14-2008

You're always welcome!