GB2416875A

GB2416875A - Switching from master computer system to standby computer system

Info

Publication number: GB2416875A
Application number: GB0417215A
Authority: GB
Inventors: Oliver Veits
Original assignee: Siemens AG
Current assignee: Siemens AG
Priority date: 2004-08-03
Filing date: 2004-08-03
Publication date: 2006-02-08
Also published as: GB0417215D0; WO2006012968A1

Abstract

A method of switching over from a master computer system MR1 (such as a master router) to a standby computer system BR1 (such as a standby router) via an intermediate switching system L2 Switch where control messages are intermediately stored together with the identification of the port the control messages were received from, a link state of that port is detected, and, in the case of a link down condition, the messages are set down to a lower priority level and are sent to a standby computer. The system is particularly applicable to routers. The standby router instantaneously assumes the active role if the master router fails in a way that it has no opportunity to send a further "hello" message with priority 0. The Virtual Router Redundancy Protocol (VRRP) or the Hot Standby Router Protocol (HSRP) may be used. The switching system may comprise OSI-Layer-2 functions.

Description

24 1 6875 Method and switching system for switching over from a master

computer system to a standby computer system

Description

The present invention relates to a method and switching system for switching over from a master computer system to a standby computer system

BACKROUND OF INVENTION

This invention may particularly be used in a LAN environment.

In IP Ethernet networks simple terminals such as PCs or telephones for example as a rule have a single static gateway entry in the form of an IP address (IP address of the router). If the terminals wish to send IP packets to IP addresses which lie outside their own sub network, they must send the IP packets in the Ethernet environment (also: Token Ring, FDDI) to the MAC address of the gateway. So that a failure of the gateway does not result in the terminals becoming completely decoupled from the rest of the world, redundant gateways are used which behave as a single gateway from the standpoint of the terminal. For this to be done the two gateways must agree as to which one assumes the active role. The associated protocols VRRP (= Virtual Router Redundancy Protocol) and HSRP (= Hot Standby Router Protocol) are described in the two Request for comments documents RFC 2338 (IETF Network Working Group: S. Knight, D. Weaver Ascend Communications, Inc., D. Whipple Microsoft, Inc., R. Hinden, D. Mitzel, P. Hunt Nokia, P. Higginson, M. Shand Digital Equipment Corp., A. Lindem IBM Corporation, April 1998) and RFC 2281 (IETF Network Working Group: T. Li, B. Cole Juniper Networks, P. Morton, D. Li Cisco systems, March 1998, http://www.rfc- editor.org/rfc.html).

The active router (computer system) sends out "hello" control messages periodically. The standby router (computer system) listens to the "hello" control messages. If a certain number of "hellos control messages do not arrive at the standby router, the standby router works on the assumption that the active router is no longer present and takes over the role of the active router. If with HSRP the times specified in RFC2281 are adhered to then the switchover time on failure of the active router is 10 seconds (recommended Holdtime). With VRRP the value is slightly over 3 seconds (more precisely (3* advertisement_interval) + Skew_time; default for advertisement_interval is one second, Skew time is (256 - priority)/256, which means it is between 1/256 second and 1 second).

SUMMARY OF INVENTION

The present invention aims to minimize the service downtime, if the active router fails.

Said problem is solved by the features mentioned in the independent claims. Preferred embodiments of the invention are described in the dependent claims.

A main aspect of the invention is a method to switch over from a master computer system to a standby computer system via an intermediate switching system comprising the steps of: receiving control messages from the master computer (MRl) system, detecting the port identification (ID), where the control message has been received from, - intermediately storing the control messages, together with the port identification information, detecting the link state of the corresponding port and sending stored control messages set down to a lower priority level to the standby computer system (BRl) in the case of a detected link down condition on the port identification stored in memory.

The stored control messages can be passed through without any changes to the standby computer (BRl) system in the case that no link down condition is detected on the stored ingress port ID.

A further aspect of the invention is a switching system to perform the above mentioned method comprising: means for receiving control messages from the master computer system (MRl), means for detecting the port identification, where the control message has been received from, means for intermediately storing the control messages, together with the port identification information, means for detecting the link state of the corresponding port, means for passing through the stored control messages to the standby computer system (BRl), means for setting down the priority level of the stored control messages prior to passing them to the standby computer system (BRl) in the case of detected link down condition on the port identification stored in the memory and - means for setting down the priority level of the stored control messages and sending the manipulated control messages to the standby computer system/ triggered by the detection of a link down condition on the previously stored ingress port identification, caused by a master computer system's failure or of the failure of the next-hop network device in the direction to the master computer system.

The proposed invention provides the following advantages: The standby computer system instantaneously assumes the active role if the Master Router or its Ethernet network card is defective and it thus has no further opportunity of sending a "hello" control message with a priority of 0. If the active computer system fails, the service downtime is minimized from a range of a few seconds to a value of a few milliseconds.

BRIEF DESCRIPTION OF THE DRAWINGS

Preferred embodiments of the present invention will now be described with reference to the accompanying drawing in which: The Figure is a schematic network diagram from RFC2238.

DETAILLED DESCRIPTION OF THE DRAWINGS

The Figure depicts Host computers Hl, H2, H3, H4 which are connected via default routers IP A to a OSI-Layer-2 switching system L2 Switch. Furthermore there is a Master router (computer system) MRl and a Standby or Backup Router (computer system) BRl which are both connected to the switching system L2 Switch. The "star" nearby the routers MRl and BRl signifies the IP Address, e.g. IP A, IP B. If the master router MRl fails it loses its links to the Layer-2 switching system. In this case the Layer 2 switching system must ensure that the backup router is not waiting for the timeout specified above before assuming the active role.

To this end the L2 switching system sends a "hello" control message to the Backup router BRl with the same contents as the periodic "hello" control messages of Master Router MRl, but with the priority 0. With VRRP priorities 0 has a special significance it indicates that the master router withdraws in an orderly manner. After receiving packets of this type the Backup Router immediately assumes the active role and must not wait for a timeout. With HSRP the priority 0 does not have the same significance and it is necessary for the Standby router to operate in "pre-empt" mode. In "pre-empt" mode the Standby router assumes the active role if it notices that the previously active router has been given a lower priority than it has itself.

Note: In Layer 2 switching system equipped with this capability must read the "hello" control messages sent by the Master Router or Computer System and store their contents in its memory. It must also detect and store a record of the Ethernet port ID of the receiving Ethernet port (in most cases this will be the port. The Master router is attached to; however, other topologies with intermediate Layer 2 switching systems between ingress port and master Router are also possible) As an example, how to determine the associated ingress port ID, the Layer 2 switching system can read the source MAC address of the "hellos control message frame and map it to the Ethernet port ID by looking up the MAC-address in the MAC address table, and store the port ID information found there in memory. Another possibility is, that the Layer 2 switching system prepends a label, which contains information of the ingress port ID, when receiving the frame. As long as the Layer 2 switching system detects, that the links state of the stored port ID is up, then all received hellos are flooded to all other ports of the same VLAN unchanged. However, if the Layer 2 switching detects a transition of the port state to "link down", it immediately sends a "hello" control message to all other ports of the same VLAN, but with priority set to zero (0). The Backup router receives this "hello" control frame and immediately takes over the service.

Claims

Claims 1. A method to switch over from a master computer system (MR1) to a

standby computer system (BR1) via an intermediate switching system (L2 Switch) comprising the steps of: - receiving control messages from the master computer (MR1) system, detecting the port identification, where the control message has been received from, intermediately storing the control messages, together with the port identification information, detecting the link state of the corresponding port and sending stored control messages set down to a lower priority level to the standby computer system (BR1) in the case of a detected link down condition on the port identification stored in memory.
2. A method as claimed in the preceding claim wherein the stored control messages are passed through without any changes to the standby computer (BR1) system in the case that no link down condition is detected on the stored ingress port identification.
3. A method as claimed in any preceding claim wherein the standby computer (BR1) system is activated depending on the priority level of the received passed control messages.
4. A method as claimed in any preceding claim when used in a LAN environment.
5. A method as claimed in any preceding claim wherein the Virtual Router Redundancy Protocol is employed.
6. A method as claimed in any preceding claim wherein the so called Hot Standby Router Protocol is employed.
7. A method as claimed in any preceding claim wherein the intermediate switching system (L2 switch) comprises OSI- Layer-2 functions.
8. A switching system (L2 switch) to perform the method as claimed in any preceding claim comprising: - means for receiving control messages from the master computer system (MRl), - means for detecting the port identification, where the control message has been received from, - means for intermediately storing the control messages, together with the port identification information, - means for detecting the link state of the corresponding port, - means for passing through the stored control messages to the standby computer system (BRl), - means for setting down the priority level of the stored control messages prior to passing them to the standby computer system (BRl) in the case of detected link down condition on the port identification stored in the memory and - means for setting down the priority level of the stored control messages and sending the manipulated control messages to the standby computer system, triggered by the detection of a link down condition on the previously stored ingress port identification, caused by a master computer system's failure or of the failure of the next-hop network device in the direction to the master computer system.
9. A switching system as claimed in any preceding claim comprising OSILayer-2 function means.
10. A switching system as claimed in any preceding claim when used in a LAN environment.