1 / 26

Hot Sticky Random Multipath or Energy Pooling

Jon Crowcroft, http://www.cl.cam.ac.uk/~jac22. Hot Sticky Random Multipath or Energy Pooling. Multipath: Network & End2End. High level argument is simply a Trilogy rerun: Multipath is resource pooling in capacity terms Multipath IP spreads traffic

timothy
Download Presentation

Hot Sticky Random Multipath or Energy Pooling

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Jon Crowcroft, http://www.cl.cam.ac.uk/~jac22 Hot Sticky Random Multipath or Energy Pooling

  2. Multipath: Network & End2End High level argument is simply a Trilogy rerun: Multipath is resource pooling in capacity terms Multipath IP spreads traffic Currently only ECMP. pre-computed/designated equivalent paths Better would be random j out of k shortes paths Basis: stick random (DAR), per Gibbens/Kelly/Key et al Multipath TCP splits load Goal is same rate as std TCP, but use eqn to determine proportion on subpath Subpath is indeicated easily for multihomed host(s) More tricky for single home, multiple route...tbd Rate set from loss/rtt or ECN, ideally

  3. S/Congestion/Carbon Let us set a price for power in the routers and transmission lines Then we give feedback (EECN?) based on this price Time of day variation in traffic means that at night, typically we see night:daytime traffic matrix about 6 fold reduction in load So re-confugure the notional cost of the shortest paths only to be 6-10 times the price And reduce the actual rate on all links by the same ratio Need energy proportional interfaces (of course) i.e. need line cards/transmission that can Slow down and use less energy That isn't rocket science (phones and laptops do this routinely with typically 3 settings)

  4. Sprint 24 hour coast-coast trace

  5. Cue:Optimization-based congestion control (ack: uMass for slides) Resource allocation as optimization problem: how to allocate resources (e.g., bandwidth) to optimize some objective function maybe not possible that optimality exactly obtained but… optimization framework as means to explicitly steer network towards desirable operating point practical congestion control as distributed asynchronous implementations of optimization algorithm systematic approach towards protocol design

  6. Model Network: Links l each of capacity cl Sources s:(L(s), Us(xs)) L(s) - links used by source s Us(xs) - utility if source rate = xs example utility function for elastic application Us(xs) xs x1 c1 c2 x2 x3

  7. Optimization Problem maximize system utility (note: all sources “equal) constraint: bandwidth used less than capacity centralized solution to optimization impractical must know all utility functions impractical for large number of sources we’ll see: congestion control as distributed asynchronous algorithms to solve this problem “system” problem

  8. The user view user can choose amount to pay per unit time, ws Would like allocated bandwidth, xs in proportion to ws cost user’s utility • pscould be viewed as charge per unit flow for user s user problem

  9. The network view suppose network knows vector {ws}, chosen by users network wants to maximize logarithmic utility function network problem

  10. Solution existence • There exist prices, ps, source rates, xs, and amount-to-pay-per-unit-time, ws =psxs such that • {Ws} solves user problem • {Xs} solves the network problem • {Xs} is the unique solution to the system problem

  11. Proportional Fairness • Vector of rates, {Xs}, proportionally fair if feasible and for any other feasible vector {Xs*}: • result: if wr=1, then {Xs} solves the network problem IFF it is proportionally fair • Related results exist for the case that wr not equal 1.

  12. Solving the network problem Results so far: existence - solution exists with given properties How to compute solution? ideally: distributed solution easily embodied in protocol insight into existing protocol

  13. Solving the network problem • congestion “signal”: function of aggregate rate at link l, fed back to s. linear increase multiplicative decrease change in bandwidth allocation at s where

  14. Solving the network problem • Results: • * converges to solution of “relaxation” of network problem • xs(t)pl(t) converges to ws • Interpretation: TCP-like algorithm to iteratively solve optimal rate allocation!

  15. Optimization-based congestion control: summary bandwidth allocation as optimization problem: practical congestion control (TCP) as distributed asynchronous implementations of optimization algorithm optimization framework as means to explicitly steer network towards desirable operating point systematic approach towards protocol design

  16. Motivation Congestion Control: maximize user utility Traffic Engineering: minimize network congestion Given routing Rlihow to adapt end rate xi? Given traffic xihow to perform routing Rli?

  17. Congestion Control Model Users are indexed by i aggregate utility Utility Ui(xi) max. ∑i Ui(xi) s.t.∑i Rlixi ≤ cl var. x capacity constraints Source rate xi Congestion control provides fair rate allocation amongst users

  18. Traffic Engineering Model Links are indexed by l aggregatecost Cost f(ul) ul = 1 min.∑l f(ul) s.t. ul =∑i Rlixi/cl var. R Link Utilization ul Traffic engineering avoids bottlenecks in the network

  19. Model of Internet Reality Congestion Control: max ∑i Ui(xi), s.t. ∑i Rlixi ≤ cl xi Rli Traffic Engineering: min ∑l f(ul), s.t. ul =∑i Rlixi/cl

  20. System Properties Convergence Does it achieve some objective? Benchmark: Utility gap between the joint system and benchmark max. ∑i Ui(xi) s.t.Rx ≤ c Var. x, R

  21. Numerical Experiments System converges Quantify the gap to optimal aggregate utility Capacity distribution: truncated Gaussian with average 100 500 points per standard deviation Access-Core Abilene Internet2

  22. Results for Abilene: f = eu_l Aggregate utility gap Gap exists Standard deviation

  23. Simulation of the joint system suggests that it is stable, but suboptimal Gap reduced if we modify f Backward Compatible Design f(ul) Cost f f(ul) ul =1 0 Link load ul

  24. Abilene Continued: f = n(ul)n Aggregate utility gap n Gap shrinks with larger n

  25. Theoretical Results Modify congestion control to approximate the capacity constraint with a penalty function Theorem: modified joint system model converges if Ui’’(xi)≤-Ui’(xi)/xi Master Problem: min. g(x,R) = - ∑iUi(xi) + γ∑lf(ul) Congestion Control: argminx g(x,R) Traffic Engineering: argminR g(x,R)

  26. Now re-do for Energy… • Need energy propotional interfaces from vendor • Need k-shortest path (& BGP equiv) • Traffic Manager report aggregate demand • Configure rates/energy based on aggregate demand • Future:- • put topology&energy* figures (e.g. from JANET) in Xen/VM based emulator • See how much we can gain…

More Related