Question: The next slide supposedly shows an example of a blocking network. However, in the network shown, there does seem to be a path connecting any permutation of sources and destinations. Why, then, is it blocking? The reason given on the slide seems to have no relation to whether or connecting any permutation of sources and destinations is possible.
This comment was marked helpful 0 times.
tchitten
As shown in the slide, you can't actually connect 1 and 6 simultaneously with 3 and 7 since they would be forced to share a line. Thus any permutation that paired 1 with 6 and 3 with 7 would have a conflict. The question of blocking and non-blocking is whether every permutation can communicate simultaneously, not just if they can be connected.
This comment was marked helpful 0 times.
tcz
To clarify, by a "connecting any permutation of sources and destinations", this slide means the network is able to support connecting every possible configuration that connects every source node to a destination node simultaneously.
This comment was marked helpful 0 times.
eatnow
Is blocking/nonblocking a property of the topography, or the actual hardware? I.e. if the topography allows for 2 non conflicting paths, but static routing causes connections to conflict, is that considered as blocking or nonblocking?
This comment was marked helpful 0 times.
crs
I think blocking/nonblocking is a property of topology at least in the context of this and the next slides. We can determine whether a design is blocking or not by just looking at the graph while routing is more an implementation level issue.
This comment was marked helpful 0 times.
uhkiv
Could someone explain how exactly we define "cutting the network in half" in bisection bandwidth? Is it any partition of the graph into two sets?
This comment was marked helpful 0 times.
jhhardin
I believe the partition has to be into two equal sets, and generally we care about the minimum such cut where we see the worst case bandwidth. The sets are equal because we want to maximize the amount of data that will have to pass through this cut (in this case we expect half) so that we can get a meaningful idea of worst case network capacity.
This comment was marked helpful 0 times.
elemental03
I didn't really understand the term bisection bandwidth very well so i trying to find some good sources online. Here's one that i found helpful : "A bisection of a network is a partition into two equally-sized sets of nodes. The sum of the capacities of links between the two partitions is called the bandwidth of the bisection. The bisection bandwidth of a network is the minimum such bandwidth along all possible bisections. Therefore, bisection bandwidth can be thought of as a measure of worst-case network capacity"
And here is the article where i found it on; it discusses bisection bandwidth in detail and provides a great explanation using clear diagrams :
Question: The next slide supposedly shows an example of a blocking network. However, in the network shown, there does seem to be a path connecting any permutation of sources and destinations. Why, then, is it blocking? The reason given on the slide seems to have no relation to whether or connecting any permutation of sources and destinations is possible.
This comment was marked helpful 0 times.
As shown in the slide, you can't actually connect 1 and 6 simultaneously with 3 and 7 since they would be forced to share a line. Thus any permutation that paired 1 with 6 and 3 with 7 would have a conflict. The question of blocking and non-blocking is whether every permutation can communicate simultaneously, not just if they can be connected.
This comment was marked helpful 0 times.
To clarify, by a "connecting any permutation of sources and destinations", this slide means the network is able to support connecting every possible configuration that connects every source node to a destination node simultaneously.
This comment was marked helpful 0 times.
Is blocking/nonblocking a property of the topography, or the actual hardware? I.e. if the topography allows for 2 non conflicting paths, but static routing causes connections to conflict, is that considered as blocking or nonblocking?
This comment was marked helpful 0 times.
I think blocking/nonblocking is a property of topology at least in the context of this and the next slides. We can determine whether a design is blocking or not by just looking at the graph while routing is more an implementation level issue.
This comment was marked helpful 0 times.
Could someone explain how exactly we define "cutting the network in half" in bisection bandwidth? Is it any partition of the graph into two sets?
This comment was marked helpful 0 times.
I believe the partition has to be into two equal sets, and generally we care about the minimum such cut where we see the worst case bandwidth. The sets are equal because we want to maximize the amount of data that will have to pass through this cut (in this case we expect half) so that we can get a meaningful idea of worst case network capacity.
This comment was marked helpful 0 times.
I didn't really understand the term bisection bandwidth very well so i trying to find some good sources online. Here's one that i found helpful : "A bisection of a network is a partition into two equally-sized sets of nodes. The sum of the capacities of links between the two partitions is called the bandwidth of the bisection. The bisection bandwidth of a network is the minimum such bandwidth along all possible bisections. Therefore, bisection bandwidth can be thought of as a measure of worst-case network capacity"
And here is the article where i found it on; it discusses bisection bandwidth in detail and provides a great explanation using clear diagrams :
http://etherealmind.com/bisectional-bandwidth-l2mp-trill-bridges-design-value/
This comment was marked helpful 1 times.