Cisco DS-TE Lab Trial Bill Cerveny <cerveny@internet2.edu> - - PowerPoint PPT Presentation

IETF TEWG Experience Report - March 22, 2001

Cisco DS-TE Lab Trial

Bill Cerveny <cerveny@internet2.edu> John Moore <jhm@ncstate.net> Paul Schopis <pschopis@itecohio.org> and mostly as a proxy... Ben Teitelbaum <ben@internet2.edu>

IETF TEWG Experience Report - March 22, 2001

Background

• Abilene backbone trying to implement a

QBone Premium Service domain

• Service goal: EF edge-to-edge "virtual trunk"
• Cisco suggested that DS-TE might be helpful

tool to complement MDRR, CAR, QPPB,...

• Learned enough to pique our interest
• Cisco and Internet2 agree to perform EFT
• Test suite developed and executed in lab by

Internet2 ITECs

IETF TEWG Experience Report - March 22, 2001

Abilene Topology

January, 2001

• 47 connectors
• 183 participants
• 34 connections to 20 peer networks

IETF TEWG Experience Report - March 22, 2001

Early Hopes for DS-TE

• Before we were fully briefed, we thought

DS-TE might give us...

– Resource accounting / admissions control – Up-calls on constraint failure – Delay constraints too – "Trunk-style" classification and policing – Mechanism to protect PQ resources from non- policed connectors

• After briefing, understood that DS-TE only

addressed first item above

IETF TEWG Experience Report - March 22, 2001

Implementation vs. Drafts

• Support for two Class-Types (CT0, CT1)
• CT1 bounded by "Max Reservable CT1"

and by "Max Reservable Aggregate"

• CT0 bounded by "Max Reservable

Aggregate" only

• Class-Types called "bandwidth pools"
• Pre-standard IGP extensions (e.g. use of

Cisco-proprietary TLV in OSPF)

Abilene Premium Service Example

OSU Campus 2 Campus3 UNC

BE (Class-Type 0) Tunnel (BE up to 100% on every link) EF (Class-Type 1) Tunnel (EF below 10% on every link) Guaranteed Bandwidth service 10 Mbps OSU-->UNC Real-Time

Nitro (GSR C) Semtec (GSR A)

Gunpowder (GSR B) LB = 255.1 LB = 255.3 LB = 255.2 P3/3 P3/1 P5/3 P1/3 P4/3 P1/2 P3/1 P3/0 P1/1

Smartbits 6000 Smartbits 6000

P3/2 49.1 49.2 2.2 2.1 1.1 1.2 L4 L3 L5 4.1 4.2 6.1 6.2 2.1 2.2 3.1 3.2 172.16 172.16 1.1 1.2

7200-VXR-A (ARz) 7505-B (ARy) 7505-A (ARx) Smartbits 2000 Smartbits 2000

1.2 1.1 1.1 1.2 3.1 3.2

AS 1000 10.130.x.y/24 AS 2000 10.140.x.y/24 AS 3000 10.150.x.y/24

P4/2 OC-12 OC-12 OC-12 OC-12 OC-12 OC-3 OC-3 OC-3 FE FE FE

AS 4000 10.160.x.y/24

Smartbits 2000

Test Network

Source Domain Sink Domains Abilene Pseudo-Core

IETF TEWG Experience Report - March 22, 2001

Basic MPLS TE Verification

• Verification of Basic/Aggregate MPLS TE

– Global Pool (Class-Type 0) bandwidth constraint mechanism – Route selection based on administrative weight – Re-routing on link failure – Verification of correct tunnel routing via CLI and by passing traffic

Nitro (GSR C) Semtec (GSR A)

Gunpowder (GSR B) LB = 255.1 LB = 255.3 LB = 255.2 P3/3 P3/1 P5/3 P1/3 P4/3 P1/2 P3/1 P3/0 P1/1

Smartbits 6000 Smartbits 6000

P3/2 49.1 49.2 2.2 2.1 1.1 1.2 L4 L3 L5 4.1 4.2 6.1 6.2 2.1 2.2 3.1 3.2 172.16 172.16 1.1 1.2

7200-VXR-A (ARz) 7505-B (ARy) 7505-A (ARx) Smartbits 2000 Smartbits 2000

1.2 1.1 1.1 1.2 3.1 3.2

AS 1000 10.130.x.y/24 AS 2000 10.140.x.y/24 AS 3000 10.150.x.y/24

P4/2 OC-12 OC-12 OC-12 OC-12 OC-12 OC-3 OC-3 OC-3 FE FE FE

AS 4000 10.160.x.y/24

Smartbits 2000

Test Network

Source Domain Sink Domains Abilene Pseudo-Core

IETF TEWG Experience Report - March 22, 2001

DS-TE Lab Test Results1/2

• DS-TE Extensions

– Verified sub-pool route selection based on bandwidth availability:

• e.g. Best Effort can get up to 100% of link capacity

while EF tunnels were limited to the configured sub- pool budget

– Achieves separate CAC and separate Route Computation for sub-pool tunnels – Verification of correct sub-pool (Class-Type 1) tunnel routing and forwarding

IETF TEWG Experience Report - March 22, 2001

DS-TE Lab Test Results2/2

• DS-TE Extensions (continued)

– Verified sub-pool preemption mechanism (preemption orthogonal to b/w pool) – Successfully added MDRR PQ, CAR, and QPPB to approximate complete solution

IETF TEWG Experience Report - March 22, 2001

Conclusions

• DS-TE worked as advertised
• Potentially useful tool in kit (if TE needed

for non-Premium traffic)

• Complete test report to be made public "real

soon now"™