Identifying network flows under network address translation

Patent metadata

Field	Value
Publication number	US-10097510-B2
Application number	US-201313966928-A
Country	US
Kind code	B2
Filing date	Aug 14, 2013
Priority date	Mar 14, 2013
Publication date	Oct 9, 2018
Grant date	Oct 9, 2018

Abstract

The present disclosure relates generally to the field of identifying network flows (e.g., IP flows) under network address translation. In various examples, identifying network flows (e.g., IP flows) under network address translation may be implemented in the form of methods and/or algorithms.

First claim

What is claimed is:

1. A method implemented in a computer system for correlating a network flow seen at a plurality of observation points in a network, the method comprising:
obtaining, by a first processing element at a first one of the plurality of observation points, a plurality of first packets of the network flow, each of the plurality of first packets including a first address;
extracting, by the first processing element, one or more invariant fields from each of the first packets in response to the first address being unknown;
sending, by the first processing element, the first address of each of the first packets and information of the one or more invariant fields of each of the first packets to a second processing element at a second one of the plurality of observation points in response to the first address being unknown;
receiving, by the second processing element, the first address of each of the first packets and the information of the one or more invariant fields of each of the first packets;
obtaining locally, by the second processing element, a second packet of the network flow;
extracting, by the second processing element, one or more invariant fields from the second packet, in response to a second address of the second packet being unknown;
comparing, by the second processing element, the one or more invariant fields of each of the first packets received from the first processing element with the one or more invariant fields of the second packet obtained by the second processing element in response to the second address being unknown;
determining, by the second processing element, a number of third packets from the plurality of first packets based on a comparison result between the one or more invariant fields of each of the first packets with the one or more invariant fields of the second packet, in response to the second address of the second packet being unknown, wherein the one or more invariant fields of each of the third packets are matched with the one or more invariant fields of the second packet within a predetermined percentage;
determining, by the second processing element, a match between the first address of each of the first packets and the second address of the second packet based on a determination of the number of third packets exceeding a predetermined number, in response to the second address of the second packet being unknown; and
notifying, by the second processing element, of the match between the first address of each of the first packets and the second address of the second packet to the first processing element in response to the second address of the second packet being unknown.

2. The method of claim 1 , wherein each invariant field comprises at least one of:
(a) a field in an IP header that is not subject to network address translation;
(b) a field in a TCP header that is not subject to network address translation;
(c) a field in an UDP header that is not subject to network address translation; and
(d) information from a data field of the IP, TCP, or UDP packet.

3. The method of claim 2 , wherein:
(a) the field in the IP header that is not subject to network address translation comprises at least one of: (i) a packet length; (ii) a header checksum; (iii) a flag; (iv) a protocol number; (v) a source IP address if only a destination IP address has undergone network address translation; and (vi) a destination IP address if only a source IP address has undergone network address translation;
(b) the field in the TCP header that is not subject to network address translation comprises at least one of: (i) a sequence number; (ii) an acknowledgement number; (iii) a window size; (iv) a checksum; (v) a flag; (vi) a source port number if only a destination port number has undergone network address translation; and (vii) a destination port number if only a source port number has undergone network address translation;
(c) the field in the UDP header that is not subject to network address translation comprises at least one of: (i) a length; (ii) a checksum; (iii) a source port number if only the destination port number has undergone network address translation; and (iv) a destination port number if only a source port number has undergone network address translation; and
(d) the information from the data field of the IP, TCP, or UDP packet comprises a signature of the data field.

4. The method of claim 1 , further comprising:
storing, by at least one of the first and second processing elements, mapping information between the first address and the second address.

5. The method of claim 1 , further comprising:
stopping, by the first processing element, sending of one or more invariant fields associated with a fourth packet belonging to the network flow to the second processing element.

Assignees

• IBM

Inventors

• Agrawal Dakshi
• Ko Bong Jun
• Le Thai Franck
• Nicholson Robert B
• Pappas Vasileios
• Verma Dinesh C

Classifications

• H04L61/2514 Primary

  between local and global IP addresses · CPC title