Prevention and detection of IP identification wraparound errors

• US 6,894,976 B1
• Filed: 06/15/2000
• Issued: 05/17/2005
• Est. Priority Date: 06/15/2000
• Status: Active Grant

• Current Assignee: NetApp, Inc.
• Original Assignee: Network Appliance Incorporated (NetApp, Inc.)
• Inventors: Banga, Gaurav, Smith, Mark, Bots, Henk J.
• Primary Examiner(s): Yao, Kwang Bin
• Application Number:
  US09/595,599
• Time in Patent Office: 1,797 Days
• Field of Search: 370/235, 370/236, 370/241, 370/242, 370/244, 370/250, 370/395.1, 370/395.2, 370/395.21, 370/395.51, 370/395.52, 370/466, 370/468, 370/469, 370/473, 370/474
• US Class Current: 370/235
• CPC Class Codes: H04L 69/16 Implementation or adaptation of Internet protocol [IP], of transmission control protocol [TCP] or of user datagram protocol [UDP]
• CPC Class Codes: H04L 69/166 IP fragmentation; TCP segmentation

Abstract

A method of generating IP identification numbers for IP datagrams. A plurality of IP identification number generators are maintained. A plurality of receiving stations are associated with the plurality of IP identification number generators such that each receiving station has an IP identification number generator associated therewith. An IP identification number is generated for a datagram sent to a receiving station based on an output of the associated IP identification number generator. Also, a method of reducing a likelihood of misassembly of received data fragments from fragmented IP datagrams. The data fragments have an IP identification number. All received data fragments of the datagram are discarded upon detection of receipt of an overlapping data fragment having the IP identification number. Timeouts for reassembling datagrams are reduced. First, timeouts for reassembling the datagrams are reduced to less than a standard timeout. Second, timeouts are reduced upon detection of a gap in the received data fragments. Third, timeouts are reduced upon detection of a datagram having the same protocol identification number and source address, but having a different IP identification number. Furthermore, a method of detecting a likelihood of misassembly of data fragments from fragmented IP datagrams. Communication errors between a sending station and a receiving station are detected. A rate is determined at which an IP identification number generator associated with the receiving station wraps around. NFS re-transmissions are determined. Based on one or more of these, a likelihood of misassembly at the receiving station is determined to be high.

Claims

1. A method of reducing a likelihood of misassembly of data fragments from fragmented IP datagrams, comprising the steps of:

receiving data fragments of a datagram having an IP identification number; and

discarding all received data fragments of the datagram upon detection of receipt of an overlapping data fragment having the IP identification number, wherein the overlapping data fragment overlaps data in an already-received data fragment.

2. A method as in claim 1, wherein the overlapping data fragment overlaps all of the already-received data fragment.

3. A method as in claim 1, wherein the overlapping data fragment overlaps less than all of the already-received data fragment.

4. A method as in claim 1, wherein the steps are performed by an IP layer of a receiving station'"'"'s communication system.

5. A method of reducing a likelihood of misassembly of data fragments from fragmented IP datagrams, comprising the step of dynamically reducing a timeout for reassembling the datagrams to less than a standard timeout based on NFS data for round-trip times between a sending station and a receiving station.

6. A method as in claim 5, wherein the data fragment reassembly timeout is reduced to 45 seconds from the standard timeout of 64 seconds.

7. A method as in claim 5, wherein the step is performed by an IP layer of a receiving station'"'"'s communication system.

8. A method of reducing a likelihood of misassembly of data fragments from fragmented IP datagrams, comprising the steps of:

receiving data fragments of a datagram having an IP identification number; and

reducing a remaining time for reassembling the datagram upon detection of a gap in the received data fragments.

9. A method as in claim 8, wherein the remaining time for reassembling the datagram is reduced to eight seconds.

10. A method as in claim 8, wherein the steps are performed by an IP layer of a receiving station'"'"'s communication system.

11. A method of reducing a likelihood of misassembly of data fragments from fragmented IP datagrams, comprising the steps of:

receiving data fragments of a first datagram having a protocol identification number, a source address, and a first IP identification number; and

reducing a remaining time for reassembling the first datagram upon detection, before receipt of a last data fragment of the first datagram, of a data fragment of a second datagram having the protocol identification number and the source address but having a second IP identification number.

12. A method as in claim 11, wherein the remaining time for reassembling the first datagram is reduced to eight seconds.

13. A method as in claim 11, wherein the steps are performed by an IP layer of a receiving station'"'"'s communication system.