Error detection/correction and fault detection/recovery – Data processing system error or fault handling – Reliability and availability
Reexamination Certificate
1998-04-14
2001-01-09
Beausoliel, Jr., Robert W. (Department: 2785)
Error detection/correction and fault detection/recovery
Data processing system error or fault handling
Reliability and availability
C709S223000
Reexamination Certificate
active
06173411
ABSTRACT:
TECHNICAL FIELD
The present invention relates, in general, to fault-tolerant computing. More specifically, the present invention relates to methods and systems for quickly switching between network connections.
BACKGROUND OF THE INVENTION
The reliability of computer based applications continues to be an important consideration. Moreover, the distribution of applications across multiple computers, connected by a network, only complicates overall system reliability issues. One critical concern is the reliability of the network connecting the multiple computers. Accordingly, fault-tolerant networks have emerged as a solution to insure computer connection reliability.
In many applications, the connection between a single computer and a network is a critical point of failure. That is, often a computer is connected to a network by a single physical connection. Thus, if that connection were to break, all connectivity to and from the particular computer would be lost. Multiple connections from a single computer to a network have therefore been implemented, but not without problems.
Turning to
FIG. 1
, a diagram of a computer
11
connected to a network
21
is shown. Computer
11
includes a network interface, for example, a fast-Ethernet interface
13
. A connection
30
links fast-Ethernet interface
13
with a fault-tolerant transceiver
15
. Fault tolerant transceiver
15
establishes a connection between connection
30
and one of two connections
29
and
31
to respective fast-Ethernet switches
19
and
17
(these “switches” as used herein are SNMP managed network Switches). Switches
17
and
19
are connected in a fault-tolerant matter to network
21
through connections
23
and
25
.
Fault-tolerant transceiver
15
may be purchased from a number of vendors including, for example, a Digi brand, model MIL-240TX redundant port selector; while fast-Ethernet switches
17
and
19
may also be purchased from a number of vendors and may include, for example, a Cisco brand, model 5000 series fast-Ethernet switch.
Operationally, traffic normally passes from fast-Ethernet interface
13
through fault-tolerant transceiver
15
, and over a primary connection
29
or
31
to respective switch
17
or
19
and on to network
21
. The other of connections
29
and
31
remains inactive. Network
21
and switches
17
and
19
maintain routing information that directs traffic bound for computer
11
through the above-described primary route.
In the event of a network connection failure, fault-tolerant transceiver
15
will switch traffic to the other of connection
29
and
31
. For example, if the primary connection was
31
, and connection
31
broke, fault-tolerant transceiver
15
would switch traffic to connection
29
.
When, for example, traffic from computer
11
begins passing over its new, backup connection
29
through switch
19
, network routing has to be reconstructed such that traffic bound for computer
11
is routed by the network to the port on switch
19
that connection
29
is attached to. Previously, the routing directed this traffic to the port on switch
17
that connection
31
was attached to.
Several problems arise from the above-described operation. First, the rebuilding of network routing to accommodate passing traffic over the back-up connection may take an extended period of time. This time may range from seconds to minutes, depending upon factors including network equipment design and where the fault occurs. Second, fault-tolerant transceiver
15
is only sensitive to a loss of the physical receive signal on the wire pair from the switches (e.g.,
17
and
19
) to the transceivers. It is not sensitive to a break in the separate wire pair from the transceiver to the switch. Also, it is sensitive only to the signal from the switch to which it is directly attached and does not test the backup link for latent failures which would prevent a successful recovery. This technique also fails to test the switches themselves.
Another example of a previous technique for connecting a computer
11
to a network
21
is shown in FIG.
2
. Network switches
17
and
19
and their connection to each other and network
21
is similar to that shown in FIG.
1
. However, in this configuration, each of switches (e.g.,
17
and
19
) connects to its own fast-Ethernet interface (e.g.,
13
and
14
) within computer
11
.
Operationally, only one of interfaces
13
and
14
is maintained active at any time. When physical signal is lost to the active interface, use of the interface with the failed connection is ceased, and connectivity begins through the other, backup interface. The backup interface assumes the addressing of the primary interface and begins communications. Unfortunately, this technique shares the same deficiencies with that depicted in FIG.
1
. Rerouting can take an extended period of time, and the only failure mode that may be detected is that of a hard, physical connection failure from the switch to the transceiver.
The present invention is directed toward solutions to the above-identified problems.
SUMMARY OF THE INVENTION
Briefly summarized, in a first aspect, the present invention includes a method for managing network routing in a system including a first node, a second node and a third node. The first node has primary and secondary connections to the second and third nodes, respectively. Also, the second and third nodes are connected by a network.
The method includes periodically communicating between the first and the second or third node over at least the primary connection. A status of network connectivity between the communicating nodes is thereby determined.
If the network connectivity determined is unacceptable, roles of the primary and secondary connections are swapped to establish new primary and secondary connections. A message is then sent with an origin address of the first node to the second node over the new primary connection. The origin address of this message facilitates the network nodes learning about routing to the first node over the new primary connection.
As an enhancement, the first node may include a first port connected to the primary connection and a second port connected to the secondary connection. The first and second ports have first and second network addresses, respectively; and the first node has a system network address. The periodic communication may be transmitted from the first port of the first node with an origin address of the first port. Further, the origin address of the message sent if network connectivity was unacceptable may be the system network address of the first node. Also, the periodic communication may be a ping message having the first network address of the first port as its origin address. This ping message may be destined for the second or third node.
If the ping message fails, another ping message may be sent from the second port to the other of the second and third nodes, not previously pinged. If this ping message is successful, the method may include swapping the roles of the primary and secondary connections and pinging the second node over the new primary link.
As yet another enhancement, the status of the connection between the second port and the other of the second and third nodes to which the previous ping was sent is determined.
In another aspect, the present invention includes a system for implementing methods corresponding to those described hereandabove. In this embodiment a link manager may be attached to the computer and may provide connectivity between the computer and the primary and secondary connections. As implementation options, the link manager may be, for example, integral with the computer (e.g., on a main board of the computer), on an expansion board of the computer, or external to the computer. Also, the computer may be an operator workstation or a controller such as, for example, an industrial or environmental controller.
REFERENCES:
patent: 4692918 (1987-09-01), Elliott et al.
patent: 4710926 (1987-12-01), Brown et al.
patent: 4787082 (1988-11-01), Delaney et al.
patent: 4
Cummings Gene A.
Gale Alan A.
Hirst Michael D.
Barron David
Beausoliel, Jr. Robert W.
Detweiler Sean
Elisca Pierre Eddy
Morris J. J.
LandOfFree
Method and system for fault-tolerant network connection... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and system for fault-tolerant network connection..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and system for fault-tolerant network connection... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2436050