Router topology having N on 1 redundancy

Details Router topology having N on 1 redundancy Router topology having N on 1 redundancy

2001-04-20

2003-12-02

Perveen, Rehana (Department: 2182)

Electrical computers and digital data processing systems: input/

Input/output data processing

Input/output expansion

C714S001000, C714S002000, C370S389000, C370S401000, C370S412000

Reexamination Certificate

active

06658494

ABSTRACT:

FIELD OF THE INVENTION
The present invention is directed to a router topology having redundancy, and, more particularly, a router topology that decreases the number of outputs affected in the event of any single point failure in a crosspoint or crossbar module.
BACKGROUND OF THE INVENTION
Circuit switching has long been used to permit the shared use of various resources, such as cameras, tape recorders, disk storage, or special effects generators, among a number of users. Other equipment may also be connected with switched circuits in order to produce and distribute the necessary signals associated with broadcasting a television signal, producing a movie, developing a commercial, or other similar activities. The majority of circuit switching is carried out with crosspoint or crossbar matrices similar to those used by telecommunications providers for connecting telephone calls, or switching higher bandwidth consolidated telephone signals.
In a broadcast facility, it is very important that “on-air” signals do not fail. Traditionally, in order to reach this goal, a router system had to be fully redundant.
FIG. 1
illustrates a redundant router system according to the prior art. The router system
10
is an M input by N output router. It is actually configured as two M by N routers
12
, followed by N, 2 input by 1 output switches
14
. The corresponding input of each router
12
is fed the same signal, and the N outputs are likewise connected in corresponding order to the N, 2×1 switches
14
. An input distribution component formed by fan-out amplifiers
16
is used to distribute each input to each M×N switch
12
. Within each router
12
there may be multiple crosspoint or crossbar matrices, input modules, and output modules in any of a number of known configurations providing the necessary switch dimensions. The switch
12
can also be used with port oriented routers whereby input-output circuit pairs share a common connector and port interface to the switch without any loss of generality. The output switches
14
normally select outputs from only one of the routers
12
to appear at their respective outputs, however, if there is a failure in the router the output switches
14
are receiving signals from, the output switches
14
select output signals from the alternate router
12
to appear at their respective outputs.
An assumption is made that the 2×1 switches
14
are simpler than the M×N switches
12
and therefore more reliable. In addition, the 2×1 switches
14
are individually repairable and in no way interrelated to each other. Therefore, a single failure in the system will affect at most a single output.
As the size of the M×N core has grown larger and larger, the topology shown in
FIG. 1
becomes too complicated, large and expensive. For telecommunications purposes, the M×N core can be blocking, i.e., there are not enough crosspoint elements to guarantee that every input may be connected to at least one output at any given instant. In addition, such a switch also does not need to be able to couple any input to any subset of outputs, including all or some of the available outputs. Therefore, in order to save crosspoints and cost, these routers have become multistage matrices of 3 or more layers. Broadcast applications still demand a non-blocking structure where any input may be connected to any single output, group of outputs, or every output, and so the geometric growth in crosspoint cost and size has contributed to the decline of the approach shown in FIG.
1
.
Since fully redundant routers are space and cost prohibitive, and recognizing that routers are made of a number of modules for inputs, outputs and crosspoints, many broadcast and telecommunications installations are designed so that the effect of a single card failure is minimized without backup. For example, a broadcast facility may have a master control studio responsible for the last switching and production details of a television program prior to broadcast. This studio could have 4 unique signals from which to choose. If all of these signals were on a single input or output card in the switch matrix, and that single card failed, there would be nothing available to broadcast on air. However, if the 4 inputs were distributed such that each input was on a separate module and likewise each output was on a separate module, then should any one module fail, there would still be 3 signals left for the master control studio. While such a technique is not ideal, it is satisfactory and requires no additional router frames or modules. How well this approach works is based on the number of inputs and outputs, or ports if a port oriented router is used to distribute inputs and outputs, implemented on a given assembly. This number can be as small as 2 and as large as 32 in contemporary router designs.
One limitation with this approach, is that crosspoint technology has become very dense with thousands of crosspoints implemented in a single chip. In order to minimize total system size and cost, crosspoint building blocks of 128×32 and 256×64 are common in today's routers. Once crosspoint cards are this big, there may only be two or three in an entire system, therefore the impact block is substantial if a crosspoint card should fail. This limits the utility of the aforementioned approach to minimize failure impact without paying for full backup redundancy. For most systems, an empirical limit of 8 or 16 is considered to be the maximum allowable impact block size.
In other types of systems, an approach called N on 1 redundancy has become popular. A computer memory system uses this approach. In a simple example, two separate hard drives are required for caching and storing intermediate data so that a given software algorithm can execute correctly. In a classic scenario, two drives would be needed to back this system up in case of failure of either drive. However, if the drives could share common data input and output busses, then one drive could be used to back up either of the other two. There is now the need to be able to intelligently assign the standby drive to the address space of either primary drive. This is easily accomplished with some additional hardware and software. While backing up disk drives at a ratio of 2 to 1 is probably not very cost effective, there are many examples today where the backup ratio is much larger, i.e., 10 or 25 to 1. In these cases, the use of 1 disk, rather than 10 or 25, is very cost effective especially provided the application requires only a statistically small loss, rather than an absolute guarantee of zero loss.
There is thus a need for a simple compact and cost effective system that provides the necessary redundancy for broadcast routers.
SUMMARY OF THE INVENTION
According to a first aspect of the invention, there is provided a router having I inputs and O outputs. The router includes P input modules, N crosspoint modules, a redundant crosspoint module and M output modules. Each input module receives I/P of I inputs. Each crosspoint module receives I inputs and outputs O/N outputs per module and the redundant crosspoint module which receives I inputs and outputs O/N outputs. Each output module receives O/N outputs per module from a particular N crosspoint module and a corresponding O/N outputs from the redundant module. According to a second aspect of the present invention, there is provided router having I inputs and O outputs. The router includes N crosspoint modules, P input module. According to a crosspoint module and M output modules. Each of the N crosspoint modules has I inputs and O/N outputs. Each input module has I/P inputs and I/P outputs wherein each output of each input module is coupled to a unique input of each of N crosspoint modules so that an input received by any of P input modules in propagated to each of N crosspoint modules. Each output module has a first set of O/M inputs, and a second set of O/M inputs and O/M outputs wherein the M output modules are arranged in N groups of M/N output modules an

Affiliated with

Also associated with

Rating

Router topology having N on 1 redundancy does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Router topology having N on 1 redundancy, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Router topology having N on 1 redundancy will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3152474