Some Qo S Deployment Issues Shumon Huque University

Some Qo. S Deployment Issues Shumon Huque University of Pennsylvania MAGPI Giga. Po. P April 15 th 2002 - NSF/ITR Scalable Qo. S Workshop 1

University of Pennsylvania network Large research university in Philadelphia, PA n 22, 000 students, 4, 000 faculty, 10, 000 staff 48, 000 registered IP addresses 200 switched subnets Central routing between them and out to Internet and Internet 2 2

MAGPI Giga. Po. P An Internet 2 Giga. Po. P Value added services n n Commodity Internet transit Facilitator of regional edu/research initiatives Subscribers n UPENN, Lehigh U, Princeton U, PA county school units, J&J Pharmaceuticals 3

MAGPI Giga. Po. P (cont) External Connectivity n Internet 2 w OC-12 c POS to Abilene n Commodity Internet w UUNET: OC-3 w Cogent: Gigabit Ethernet w Yipes: Gigabit Ethernet (rate limited) w DCANet: Fast Ethernet 4

Who wants Qo. S? University researchers n n Qo. S researchers in CS department Research applications needing strict guarantees on latency, b/w, jitter etc Networking staff (Univ and giga. Po. P) n n Manage exploding b/w needs Enable new classes of applications w Eg. Vo. IP, video conferencing, streaming n Run non-mission critical traffic at lower priority w Eg. File sharing apps, dorm traffic, bulk transfers 5

What types of Qo. S? Diff. Serv in routing core and giga. Po. P Layer 2 priority (802. 1 p) in the switched portions of the campus network Mapping L 3 Qo. S to/from L 2 Qo. S Signalling and admission control? n n RSVP intra-domain? Aggregate reservations & map to Diffserv traffic class at edge? Bandwidth Broker signalling? 6

Diff. Serv Types of forwarding behavior we are most interested in: n n EF (Expedited Forwarding) BE (Best Effort - default PHB) LBE/Scavenger (eg. QBSS) ABE - low delay form 7

Interdomain Internet Qo. S Not very optimistic Some ISPs are starting to offer services Multiplicity of providers n n n Need for them to run interoperable Qo. S implementations Mechanisms to ask for Qo. S reservations across administrative domains Peering/SLA issues 8

Interdomain Internet 2 Qo. S More optimistic Typically one or a few Qo. S enabled I 2 backbone networks (eg. Abilene) Agreed upon Qo. S architecture Common set of operational practices and procedures Some provisioning procedures in place Existing demand from researchers 9

Deployment Challenges I 2 backbone is an R&E network, but. . Universities are using it to transport production traffic between them And not just traffic associated with meritorious research applications (one of the original ideas) Giga. Po. P is a production network providing access to I 2 and Commodity Internet 10

Deployment Challenges (2) So, we need to be very careful about changes we introduce to the network to facilitate Qo. S Don’t jeopardize existing production traffic 11

Router support for Qo. S Not mature or well tested Often the features are in experimental code trains, unsuitable for deployment in a production network n Marking, re-marking, policing, traffic shaping, appropriate queue scheduling disciplines etc Insufficient #queues to support large scale service differentiation Often software implementations of required queueing disciplines instead of hardware Obviously this situation will improve in the future 12

Router code support (cont) Example: Juniper routers 4. x release: n Can police DS BA’s but not much more 5. x release: n n More queue scheduling disciplines Per queue traffic shaping DSCP marking and re-marking DSCP based prioritization and forwarding w Eg. Assigning EF BA to a high priority queue n Mapping of 802. 1 p to Layer-3 Qo. S 13

Parallel Network Infrastucture Deploy parallel network infrastructure Place Qo. S enabled routers on this Researchers are happy, but. . Cost prohibitive 14

Qo. S policy issues Where does marking occur? n n n Endstations First hop routers or switches Edge routers Who’s allowed to mark? How to validate? n n Complexity of deploying policies Additional controls and checks to enforce the policies w Policy servers: COPS, bandwidth brokers etc 15

Inter-domain signalling No suitable mechanisms today for end 2 end inter-domain signalling of Qo. S reservations, call admission control Manual/static provisioning Bandwidth brokers/SIBBS work ongoing 16

What we do today To facilitate researchers doing wide-area Qo. S experimentation: Conscious effort not to impede n n Provide research labs with an uncongested path though campus/gigapop to Qo. S enabled Abilene network Make sure intervening routers don’t mark or remark DSCP code points 17

Abilene Qo. S testbed 18

Bandwidth Management Alternatives University has experienced rapidly increasing bandwidth requirements Overprovision the campus network Buy more commodity Internet bandwidth through the giga. Po. P Employ rate limiting where appropriate Employ lightweight Qo. S, eg. LBE/Scavenger 19

Endstation problems Network apps often unable to use available bandwidth because of problems on endstations: n n n Poorly designed applications, application protocols Insufficiently sized socket buffers Inefficient, insufficiently tuned network stacks Duplex mismatch MTU mismatch Having Qo. S in the network does not address this class of performance problems 20

Conclusion We’re interested in Qo. S Too early to deploy end 2 end reservation based Qo. S in many production networks Intra-domain Qo. S a near-term possibility n n Both reservation based and lightweight Vo. IP, degrading non-mission critical traffic End 2 End Inter-domain Qo. S difficult n Co-ordination, SLAs, inter-domain signalling 21