Adaptively maintaining sequence numbers on high availability peers

• US 8,363,549 B1
• Filed: 09/02/2009
• Issued: 01/29/2013
• Est. Priority Date: 09/02/2009
• Status: Active Grant

• Current Assignee: Juniper Networks Incorporated
• Original Assignee: Juniper Networks Incorporated
• Inventors: Zhu, Droplet, Tang, Jiafeng, Fang, Tianji
• Primary Examiner(s): Thier, Michael
• Assistant Examiner(s): Javaid, Jamal
• Application Number:
  US12/552,929
• Time in Patent Office: 1,245 Days
• Field of Search: 370/225, 370/235, 370/468, 370/229, 370/231, 370/328, 709/235, 709/227, 714/4.11, 714/712
• US Class Current: 370/235
• CPC Class Codes: H04L 63/0218 Distributed architectures, e.g. distributed firewalls
• CPC Class Codes: H04L 63/0236 Filtering by address, protocol, port number or service, e.g. IP-address or URL
• CPC Class Codes: H04L 63/0254 Stateful filtering
• CPC Class Codes: H04L 67/10 in which an application is distributed across nodes in the network software deployment G06F8/60; multiprogramming arrangements G06F9/46

Abstract

A primary network element of a high availability cluster periodically synchronizes sequence numbers with a backup network element of the cluster. The primary determines the periodicity of updates according to a window size and a scale factor. The backup uses probe packets after switchover to discover the current sequence number. In one example, the primary network element includes an interface to receive first and second packets of a packet flow, a control unit to determine whether a difference between sequence numbers of the first and second packets exceeds an update window size value associated with the packet flow that is based on a window size value and a window scale factor associated with the packet flow, and a backup network interface configured to send an update message comprising the second sequence number to the backup network element of the high-availability cluster.

Claims

1. A method comprising:

receiving, with a primary network element of a high-availability cluster, a first packet of a packet flow, the first packet having a first sequence number;

receiving, with the primary network element, a second packet of the packet flow, the second packet having a second sequence number;

calculating a first difference value comprising a difference between the second sequence number and the first sequence number;

after determining that the first difference value exceeds a first update window size value associated with the packet flow, wherein the first update window size value is based on a window size value associated with the packet flow and a window scale factor associated with the packet flow, sending, with the primary network element, a first update message comprising the second sequence number to a backup network element of the high-availability cluster;

receiving a modified window size value for the packet flow;

calculating a second update window size value according to the modified window size value and the window scale factor;

receiving a third packet of the packet flow, the third packet having a third sequence number;

calculating a second difference value comprising a difference between the third sequence number and the second sequence number; and

after determining that the second difference value exceeds the second update window size value, sending a second update message comprising the third sequence number to the backup network element.

2. The method of claim 1, further comprising calculating the update window size value associated with the packet flow.

3. The method of claim 2, wherein calculating the update window size value comprises multiplying the window size value by two raised to a power equal to the window scale factor.

4. The method of claim 2, wherein calculating the update window size value comprises left-shifting the window size value by the window scale factor.

5. The method of claim 1, further comprising switching over from the primary network element to the backup network element.

6. A method comprising:

receiving, with a primary network element of a high-availability cluster, a first packet of a packet flow, the first packet having a first sequence number;

receiving, with the primary network element, a second packet of the packet flow, the second packet having a second sequence number;

receiving a synchronization (SYN) packet of the packet flow, a synchronization-acknowledgement (SYN-ACK) packet associated with the packet flow, and an ACK packet of the packet flow;

retrieving a window size value and a window scale factor from at least one of the SYN packet, the SYN-ACK packet, and the ACK packet;

calculating a difference value comprising a difference between the second sequence number and the first sequence number;

after determining that the difference value exceeds an update window size value associated with the packet flow, wherein the update window size value is based on the window size value associated with the packet flow and the window scale factor associated with the packet flow, sending, with the primary network element, an update message comprising the second sequence number to a backup network element of the high-availability cluster; and

sending the window size value and the window scale factor value to the backup network element.

7. The method of claim 6, further comprising:

retrieving a maximum segment size value from at least one of the SYN packet, the SYN-ACK packet, and the ACK packet; and

sending the maximum segment size value to the backup network element.

8. A method comprising:

receiving, with a primary network element of a high-availability cluster, a first packet of a packet flow, the first packet having a first sequence number;

receiving, with the primary network element, a second packet of the packet flow, the second packet having a second sequence number;

calculating a first difference value comprising a difference between the second sequence number and the first sequence number;

after determining that the first difference value exceeds an update window size value associated with the packet flow, wherein the update window size value is based on a window size value associated with the packet flow and a window scale factor associated with the packet flow, sending, with the primary network element, a first update message comprising the second sequence number to a backup network element of the high-availability cluster;

receiving a third packet of the packet flow, the third packet having a third sequence number;

calculating a second difference value comprising a difference between the third sequence number and the second sequence number; and

after determining that the second difference value exceeds the update window size value, sending a second update message comprising the third sequence number to the backup network element.

9. A method comprising:

receiving, with a primary network element of a high-availability cluster, a first packet of a first packet flow, the first packet having a first sequence number;

receiving, with the primary network element, a second packet of the first packet flow, the second packet having a second sequence number;

calculating a first difference value comprising a difference between the second sequence number and the first sequence number;

after determining that the first difference value exceeds a first update window size value associated with the packet flow, wherein the first update window size value is based on a first window size value associated with the first packet flow and a first window scale factor associated with the first packet flow, sending, with the primary network element, a first update message comprising the second sequence number to a backup network element of the high-availability cluster;

receiving a third packet of a second packet flow, the third packet having a third sequence number;

receiving a fourth packet of the second packet flow, the fourth packet having a fourth sequence number;

calculating a second difference value comprising a difference between the fourth sequence number and the third sequence number; and

after determining that the second difference value exceeds a second update window size value associated with the second packet flow, wherein the second update window size value is based on a second window size value associated with the second packet flow and a second window scale factor associated with the second packet flow, sending a second update message comprising the fourth sequence number to the backup network element, wherein the second update message is associated with the second packet flow.

10. A primary network element of a high-availability cluster configured to operate in a cluster mode, the network element comprising:

a network interface configured to receive a first packet of a packet flow, the first packet having a first sequence number, to receive a second packet of the packet flow, the second packet having a second sequence number, to receive a modified window size value for the packet flow, and to receive a third packet of the packet flow, the third packet having a third sequence number;

a control unit configured to calculate a first difference value comprising a difference between the second sequence number and the first sequence number, to determine whether the first difference value exceeds a first update window size value associated with the packet flow, wherein the first update window size value is based on a window size value associated with the packet flow and a window scale factor associated with the packet flow, to calculate a second update window size value according to the modified window size value and the window scale factor, to calculate a second difference value comprising a difference between the third sequence number and the second sequence number, and to determine whether the second difference value exceeds the second update window size value; and

a backup network interface configured to send a first update message comprising the second sequence number to a backup network element of the high-availability cluster when the first difference value exceeds the first update window size and to send a second update message comprising the third sequence number to the backup network element when the second difference value exceeds the second update window size.

11. The primary network element of claim 10, wherein the control unit is configured to calculate the update window size value.

12. The primary network element of claim 10, wherein the primary network element comprises a primary stateful firewall, and wherein the backup network element comprises a backup stateful firewall.

13. The primary network element of claim 12, wherein the control unit is configured to perform stateful inspection of packets of the packet flow to detect network attacks in the packets and to block the packets in which the network attacks are detected.

14. The primary network element of claim 10, wherein the primary network element comprises a primary network device.

15. The primary network element of claim 10, wherein the primary network element comprises a primary control unit of a network device.

16. The primary network element of claim 10, wherein the primary network element maintains a plurality of different update window size values for a plurality of different packet flows, and wherein the primary network element computes the update window size value for each of the packet flows as a function of a window size value associated with the respective packet flow and a window scale factor defined by the primary network element for the respective packet flow.

17. A primary network element of a high-availability cluster configured to operate in a cluster mode, the network element comprising:

a computer-readable medium comprising a flow table;

a network interface configured to receive a first packet of a packet flow, the first packet having a first sequence number, to receive a second packet of the packet flow, the second packet having a second sequence number, and to receive a synchronization (SYN) packet of the packet flow, a synchronization-acknowledgement (SYN-ACK) packet associated with the packet flow, and an ACK packet of the packet flow;

a control unit configured to retrieve a window size value and a window scale factor from at least one of the SYN packet, the SYN-ACK packet, and the ACK packet and store the window size value and the window scale factor in an entry of the flow table, wherein the entry is associated with the packet flow, to calculate a difference value comprising a difference between the second sequence number and the first sequence number, and to determine whether the difference value exceeds an update window size value associated with the packet flow, wherein the update window size value is based on a window size value associated with the packet flow and a window scale factor associated with the packet flow; and

a backup network element interface configured to send the window size value and the window scale factor value to the backup network element, and to send an update message comprising the second sequence number to a backup network element of the high-availability cluster.

18. A non-transitory computer-readable storage medium encoded with instructions for causing a programmable processor of a primary network element of a high-availability cluster to:

calculate a first update window size value associated with a packet flow based on a first window size value associated with the packet flow and a window scale factor associated with the packet flow;

receive a first packet of the packet flow, the first packet having a first sequence number;

receive a second packet of the packet flow, the second packet having a second sequence number;

calculate a first difference value comprising a difference between the second sequence number and the first sequence number;

determine whether the first difference value exceeds the first update window size value;

send a first update message comprising the second sequence number to a backup network element of the high-availability cluster when the difference value is determined to exceed the update window;

receive a modified window size value for the packet flow;

calculate a second update window size value according to the modified window size value and the window scale factor;

receive a third packet of the packet flow, the third packet having a third sequence number;

calculate a second difference value comprising a difference between the third sequence number and the second sequence number; and

after determining that the second difference value exceeds the second update window size value, send a second update message comprising the third sequence number to the backup network element.

19. A method comprising:

receiving, with a backup network element of a high-availability cluster, a first update message from a primary network element of the high-availability cluster, wherein the first update message comprises a first sequence number of a packet of a packet flow, wherein a difference between the first sequence number and a previously received sequence number corresponds to a first update window size;

receiving, with the backup network element, a second update message from the primary network element, wherein the second update message comprises a second sequence number, wherein a difference between the second sequence number and the first sequence number corresponds to a second update window size;

receiving, with the backup network element, a message indicating that the backup network element is to become active and act as a primary network element;

after becoming active, sending, with the backup network element, probe packets to a client device of the packet flow and a server device of the packet flow, wherein the probe packets each comprise a probe sequence number equal to the second sequence number;

receiving, with the backup network element, at least one of a retransmission comprising a current sequence number from the client device or an acknowledgement message comprising the current sequence number from the server device; and

associating the current sequence number received from the retransmission or the acknowledgement message with the packet flow.

20. The method of claim 19, wherein the probe packets comprise a first set of probe packets, and wherein the probe sequence number comprises a first probe sequence number, the method further comprising:

receiving a maximum segment size value associated with the packet flow from the primary network element; and

sending a second set of probe packets to the client device and the server device, wherein the second set of probe packets each comprise a second probe sequence number corresponding to the sequence number of the update message incremented by a multiple of the maximum segment size value.

21. The method of claim 20, wherein receiving the retransmission or the acknowledgement message comprises receiving the retransmission or the acknowledgement message in response to the second set of probe packets.

22. The method of claim 20, further comprising:

starting a timer after sending the first set of probe packets; and

wherein sending the second set of probe packets comprises sending the second set of probe packets when the timer expires.

23. A backup network element of a high-availability cluster configured to operate in a cluster mode, the backup network element comprising:

a primary network element interface configured to receive a first update message from a primary network element of the high-availability cluster, wherein the first update message comprises a first sequence number of a packet of a packet flow, wherein a difference between the first sequence number and a previously received sequence number corresponds to a first update window size, to receive a second update message from the primary network element, wherein the second update message comprises a second sequence number, wherein a difference between the second sequence number and the first sequence number corresponds to a second update window size, and to receive an activate message indicating that the backup network element is to become active and act as a primary network element;

one or more network interfaces configured to send probe packets to a client device of the packet flow and a server device of the packet flow, wherein the probe packets each comprise a probe sequence number equal to the second sequence number, and to receive at least one of a retransmission comprising a current sequence number from the client device or an acknowledgement message comprising the current sequence number from the server device; and

a control unit configured to cause the backup network element to become active in response to the activate message and to associate the current sequence number received from the retransmission or the acknowledgement message with the packet flow.

24. The backup network element of claim 23, wherein the probe packets comprise a first set of probe packets, wherein the probe sequence number comprises a first probe sequence number,

wherein the primary network element interface is further configured to receive a maximum segment size value associated with the packet flow from the primary network element, and

wherein the one or more network interfaces are configured to send a second set of probe packets to the client device and the server device, wherein the second set of probe packets each comprise a second probe sequence number corresponding to the sequence number of the update message incremented by a multiple of the maximum segment size value.

25. The backup network element of claim 24, wherein the control unit is configured to start a timer after the first set of probe packets are sent, and wherein the one or more network interfaces are configured to send the second set of probe packets after the timer expires.

26. A non-transitory computer-readable storage medium encoded with instructions for causing a programmable processor of a backup network element of a high-availability cluster to:

receive a first update message from a primary network element of the high-availability cluster, wherein the first update message comprises a sequence number of a packet of a packet flow, wherein a difference between the first sequence number and a previously received sequence number corresponds to a first update window size;

receive a second update message from the primary network element, wherein the second update message comprises a second sequence number, wherein a difference between the second sequence number and the first sequence number corresponds to a second update window size

receive a message indicating that the backup network element is to become active and act as a primary network element;

after becoming active, send probe packets to a client device of the packet flow and a server device of the packet flow, wherein the probe packets each comprise a probe sequence number equal to the second sequence number;

receive at least one of a retransmission comprising a current sequence number from the client device or an acknowledgement message comprising the current sequence number from the server device; and

associate the current sequence number received from the retransmission or the acknowledgement message with the packet flow.

27. The non-transitory computer-readable storage medium of claim 26, wherein the probe packets comprise a first set of probe packets, and wherein the probe sequence number comprises a first probe sequence number, further comprising instructions to:

receive a maximum segment size value associated with the packet flow from the primary network element; and

send a second set of probe packets to the client device and the server device, wherein the second set of probe packets each comprise a second probe sequence number corresponding to the sequence number of the update message incremented by a multiple of the maximum segment size value.

28. The non-transitory computer-readable storage medium of claim 27, wherein the instructions to receive the retransmission or the acknowledgement message comprise instructions to receive the retransmission or the acknowledgement message in response to the second set of probe packets.

29. The non-transitory computer-readable storage medium of claim 27, further comprising instructions to:

start a timer after sending the first set of probe packets; and

wherein the instructions to send the second set of probe packets comprise instructions to send the second set of probe packets when the timer expires.

30. A system forming a high-availability cluster of network devices, the system comprising:

a primary network element comprising;

a network interface configured to receive a first packet of a packet flow, the first packet having a first sequence number, to receive a second packet of the packet flow, the second packet having a second sequence number, to receive a modified window size value for the packet flow, and to receive a third packet of the packet flow, the third packet having a third sequence number;

a first control unit configured to calculate a first difference value comprising a difference between the second sequence number and the first sequence number, to determine whether the first difference value exceeds a first update window size value associated with the packet flow, wherein the first update window size value is based on a window size value associated with the packet flow and a window scale factor associated with the packet flow, to calculate a second update window size value according to the modified window size value and the window scale factor, to calculate a second difference value comprising a difference between the third sequence number and the second sequence number, and to determine whether the second difference value exceeds the second update window size value; and

a backup network interface; and

a backup network element comprising;

a primary network element interface configured to receive a first update message from the primary network element, wherein the update message comprises the second sequence number, to receive a second update message comprising the third sequence number from the primary network element, and to receive an activate message indicating that the backup network element is to become active and act as a primary network element;

one or more network interfaces configured to send probe packets to a client device of the packet flow and a server device of the packet flow, wherein the probe packets each comprise a probe sequence number equal to the second sequence number, to receive at least one of a retransmission comprising a current sequence number from the client device or an acknowledgement message comprising the current sequence number from the server device, and

a second control unit configured to cause the backup network element to become active in response to the activate message and to associate the current sequence number received from the retransmission or the acknowledgement message with the packet flow,

wherein the backup network interface of the primary network element is configured to send the first update message comprising the second sequence number to the backup network element when the first difference value exceeds the first update window size and to send the second update message comprising the third sequence number to the backup network element when the second difference value exceeds the second update window size.