430 likes | 601 Views
S y n e r g i s t i c Network Operations. Saqib Raza University of California, Davis. A Snapshot Of Network Operations. Scheduling. Accounting. Maintenance. Firewalls. Forensics. Inter-domain TE. Power Management. Traffic Policing. Diagnostics. Intra-domain TE. Forwarding.
E N D
SynergisticNetwork Operations Saqib Raza University of California, Davis
A Snapshot Of Network Operations Scheduling Accounting Maintenance Firewalls Forensics Inter-domain TE Power Management Traffic Policing Diagnostics Intra-domain TE Forwarding Overlay Routing
Example: Inter-Operation Dynamics B A Initially, traffic between overlay nodes A and D does not traverse ISP-A D ISP A x y C ISP-A alters link weights to direct away from link (x,y). Sensing reduced delay through ISP-A the routing overlay starts sending traffic from A to D through ISP-A Intra-domain TE Overlay Routing
The Hippocratic Oath For Network Operations • Do No Harm • Operations should be cognizant of any disruptive effects to other operations. Strive to do Good Operations should seek to enhance the efficacy of other operations.
Summary/Outline • Interface-Split Forwarding for Finer-Grained Traffic Engineering [Performance `07, Eval `07] • Cooperative Peer-to-Peer Repair of 3G Broadcast Losses [Broadnets `08, ICC `08, ICME `07] • Network-level footprints of Online Social Network Applications [IMC `09, IMC `08] • Graceful Network State Migration [Infocom `09] • MeasuRouting: A Framework for Routing Assisted Traffic Monitoring [Infocom `10] • Future Directions
Maintenance Graceful Network Migration minimizing performance disruption during planned network maintenance … Joint work with: Yuanbo Zhu & Chen-Nee Chuah (UC Davis) Intra-domain TE
Motivation • Premeditated network tasks can be judiciously scheduledto minimize performance disruption
LMS: Illustrative Example Link Capacity = C 1 Link Weights Flow Size = ½ C 2 1 1 1 1 Max Link Util = 50% 3 1 I need to repair links (a,c) and (c,f) g c a e f b d Careful! Watch out for the Maximum Link Utilization (MLU)
1 1 1 1 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 100% 3 3 3 1 1 1 1 g g g g c c c c a a a a e e e e f f f f b b b b d d d d 3
1 1 1 1 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1 3 3 3 1 1 1 1 1 g g g g c c c c a a a a e e e e f f f f b b b b d d d d 3
LMS: Illustrative Example • The schedule with multiple links simultaneously deactivated causes less disruption
Specify (sinitial,sfinal), A, B, & C to define a concrete GSM problem, e.g., LMS The General GSM Problem • sn , , n r d n n r d n n r r n r r d d A A repaired deactivated not repaired
Computational Complexity GSM is a combinatorial optimization problem 002 011 001 122 101 Solution space of LMS has 2n!/2nsolutions 000 010 212 222 020 100 220 110 200
Ants Colony Optimization Swarm intelligence meta-heuristic f f f Near optimal solutions for the Traveling Salesman Problem n n n
Performance Evaluation Single-Failure Heuristic works well generally What about the worst case? > 20 node/80 link topology > 100 experiments per data point > Report Cost Reduction (MLU) over Single-Failure Heuristic
GST: Applications Link Weight Reassignment Scheduling
Outline • Graceful Network State Migration [Infocom `09] • MeasuRouting: A Framework for Routing Assisted Traffic Monitoring [Infocom `10] • Future Directions
Measurements MeasuRouting a framework for routing assisted network measurements… Joint work with: Guanyao Huang & Chen-Nee Chuah (UC Davis) Srini Seetharaman & Jatinder Singh (DT Labs) Intra-domain TE
An evolving universe The Monitor Placement Problem Oops! important very important ? ? 1. Measurement objectives change 2. New Traffic gets introduced 3. Traffic placement changes
Measurements Intra-domain TE
Congestion TE Policy Violation
Compliant Rerouting Monitor TE policy is defined for aggregated flows Sub-populations of aggregated flows, indistinguishable from a TE perspective, can be distinguishable from a measurement perspective
1. Aggregated TE Flows e.g. OD pair traffic 2. Traffic placement given: Γ(i,j)E TE Flowset (macro-flowset) • 1. TE flowset de-composes into kmeasurement flowsets • 2. A measurement flowsethas: • a) Size • b)Importance • 3. Decision variable: • (i,j)E Measurement Flowsets (micro-flowsets) 27
Maximize score across allmeasurement flowsets across alllinks MeasuRouting Objective Flowset Size Flowset Routing Network Flow Conservation Constraints Ensure that TE performance remains within some value of the default TE performance 1 2 Link Sampling Rate Flowset Importance Points gained for sampling flowset y on link (i,j)
Report improvement in Measurement Score over default routing Synthetic Experiments • Select the number of Measurement Flowsets per OD pair (K) • Divide all flows between an OD pair into the K measurement flowsets • Assign size and importance of the measurement flowsets • Choose the permissible TE violation parameter
Network Size • Performance sensitive to number of multiple paths AS1221 44 nodes AS1239 52 nodes
Degrees Of Freedom AS1221 44 nodes • Diminishing marginal returns of increasing k
A Real Application • Trace Capture for Deep Packet Inspection (DPI) • Trace capture infrastructure selectively deployed • Increase representation of interesting traffic in traces Abilene 9 nodes ln(1-|P(i)-Q(i)|)
Outline • Graceful Network State Migration [Infocom `09] • MeasuRouting: A Framework for Routing Assisted Traffic Monitoring [Infocom `10] • Future Directions
Optimal States Of Being Graceful Network State Migration
Data Center Job Scheduling Scheduling Power conserved by switching off data center components, dynamic voltage scaling etc. Power Management Jobs scheduled on different servers to optimize performance (MapReduce, Dyrad). Jointly optimize job scheduling and power management decisions.
Data Center Load Distribution Inter-domain TE Data center operation costs vary geographically due to energy market price fluctuations [Qureshi `09] Power Management Makes sense to operate data centers in diverse energy markets. Data center load can not be instantaneously shifted from one location to another. Chalk out optimal state trajectory of BGP route advertisements.
A Calculus For Synergistic Operations Revenue Contribution CPU Cycles Bandwidth Network-wide Security Power • Each marginal unitof a resource ought to be allocated to the operation that derives the highest marginal utility from consuming it.
Questions wwwcsif.cs.ucdavis.edu/~raza www.ece.ucdavis.edu/rubinet
Measurement Utility Diversity AS1221 44 nodes • Performance improves with variance in importance