Does anyone use intelligent algorithms for determining IGP costs?
It's a widely academic topic on solving a NP-Hard problem, but
I've found most providers tend to stick to simple functions
based on RTT and link capacity.
With bandwidth costs becoming more and more reasonable, it
also seems to have been a simple "get bigger pipes" between
point A and B solution.
However, with budgets being tighter, I would assume some
more thought has been put in to using more intelligent
methods of traffic optimization.
(briefly ignoring the MPLS TE "solution")
An added twist appears to be that a full demand matrix
is difficult to build without large systems of flow
collection. I'm curious if the problem can easily be
approached without full insight in to this data.
(per link usage yes, but without end-to-end flows)
I've seen a few modeling applications, but nothing that
attempts to solve the general optimization problem.
Thoughts? Is there a need for something like this?
Are there existing effective solutions?
Thanks,
Joel Krauska
Does anyone use intelligent algorithms for determining IGP costs?
snip
An added twist appears to be that a full demand matrix
is difficult to build without large systems of flow
collection. I'm curious if the problem can easily be
approached without full insight in to this data.
(per link usage yes, but without end-to-end flows)
It's not merely flow data. You also have to include variables such as cost of
the circuit(s) at a given data rate.
/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\
Patrick Greenwell
Stealthgeeks,LLC. Operations Consulting
http://www.stealthgeeks.net
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
> An added twist appears to be that a full demand matrix
> is difficult to build without large systems of flow
> collection. I'm curious if the problem can easily be
> approached without full insight in to this data.
> (per link usage yes, but without end-to-end flows)
It's not merely flow data. You also have to include variables such as
cost of
the circuit(s) at a given data rate.
True, but this even further complicates the optimization.
You could equally consider multiple upstream provider
transit costs. I was trying to keep it simple.
I was simplifying based on the fact that most large
providers own or lease their own backbones, and don't
directly pay per bit on their own network.
-Joel
An added twist appears to be that a full demand matrix
is difficult to build without large systems of flow
collection. I'm curious if the problem can easily be
approached without full insight in to this data.
(per link usage yes, but without end-to-end flows)
Methods for reducing the number of flow collection points are discussed in this paper:
http://www.research.att.com/~jrex/papers/sigcomm00.pdf
Then there's always the MPLS as measurement platform approach. Set up a mesh of LSPs and you have edge-to-edge (or perhaps POP-to-POP to reduce the number of LSPs) traffic demands. Less ambitious variations on this can be used to classify limited amounts of traffic for tactical TE.
Something like Juniper's DCU might provide enough info, if 16 buckets is granular enough.
I've seen a few modeling applications, but nothing that
attempts to solve the general optimization problem.
There have been a couple of interesting papers on the topic recently. See, for example, work by Thorup et. al. and Ben-Ameur et. al. I'm not aware of released code that actually implements these papers, though.
Bradley
For IGP metrics, I use a formula that goes like this:
m=d*(l/s)+l+10
Where m is the metric, d is the distance between the endpoints, l is the
latency, and s is the link speed (in Mbps)
Works well for us