Tiziana Ferrari Diffserv testing in the wide area TFTANT task force

1
Diffserv testing in the wide area

• TF-TANT task force
• Tiziana Ferrari
• Italian National Institute for Nuclear Physics
• ferrari_at_cnaf.infn.it

2
European research network model according to
diffserv
diffserv domain ? National Research Network
(NRN) diffserv region ? NRNs ? diffserv
core domain
(marking)
3
Diffserv test network (Nov 99)
Cabletron Ethernet sw Cisco C7200, C7500 IBM
2212, IBM 2216 Linux SmartBits 200 (Netcom)
4
Work accomplished (June 99-Oct 99)

• network configuration, baseline performance
  testing
• -1- marking
• -2- policing functionality, token bucket tuning
• -3- scheduling
• CB-WFQ
• SCFQ (Self Clocked Fair Queuing)
• -4- QoS measurement (ongoing)
• goal validation of EF PHB
• parameters one-way delay and one-way delay
  variation
• LAN and WAN
• -5- definition, implementation and validation of
  test services
• point to point virtual leased line VLL
• one-to-many virtual leased lines mVLL

5
Policing

• -1- two-parameter token bucket
• burst size tuning according to different
  policing rates
• -2- multi-parameter token bucket
• normal and exceed burst (TCP)
• -3- policing and scheduling
• token bucket size and premium queue size

6
Policing, two-parameter TBoptimum burst size
for different traffic rates
IBM 2212

• Small TCP burst tolerance it can completely
  starve a TCP stream
• optimum token bucket size function of the
  policing rate

7
Policing, multi-parameter TB normal and exceed
burst size tuning
(Target rate 1.25 Mbps)
CISCO C7200
rule exceed burst 2 normal independently
of the platform normal burst 0.5 sec R
8
Scheduling and policingtwo-parameter TB and
premium buffer size
Recommendation token bucket size lt premium
buffer size (for low packet loss probability)
9
Scheduling WFQ

• CB-WFQ (cisco) and SCFQ (IBM)
• traffic isolation between EF BA, AF BA and
  best-effort traffic
• combinations of TCP and UDP traffic
• different aggregation degrees (variable number of
  TCP streams)
• class selectors, EF and AF
• bandwidth allocation
• no starvation
• dynamic allocation among lightly loaded and
  overloaded classes

10
SchedulingCB-WFQ and traffic isolation - WAN -

• Inconsistent results, for some sites TCP
  throughput lt 1.25 Mbps (target rate)
• different ATM burst tolerances
• tuning of configuration at layer 2 needed
• shaping
• overall TCP performance improves with larger
  aggregation degrees
• traffic isolation on WAN connections can be
  an issue

11
Scheduling Self Clocked Fair Queuing - LAN -

• IBM 2212
• Policy (traffic profile, validity period,
  diffserv action)
• Diffserv action (type of marking, queue type,
  bandwidth)
• optimum traffic isolation (UDP)

Premium 163.8 Kbps guaranteed (8 PPP link
bw) Assured 819.2 Kbps (40 PPP bw)
12
Services

• Virtual leased line
• replacement of ATM permanent connections, support
  to managed bandwidth service
• QoS support in/to NRN which are not ATM based
• point to point VLL
• one to many mVLL
• capacity allocation on congested links (e.g. US
  links)
• better-than-best-effort (qualitative definition)

13
ServicesmVLL
14
ServicesmVLL implementation
15
QoS measurement 1-way delay and instantaneous
delay variation (LAN)
Goal VLL service implementation and
validation Metric Type-P-One-way-Delay (RFC
2679), type-P-one-way-ipdv Variables

• WFQ/SCFQ EF queue depth
• packet size
• EF queue weight
• traffic volume

16
One-way delay average vs packet size
1518 by
1280 by
1024 by
512 by
256 by
128 by
64 by
1248 RFC
17
Latency in time
18
ipdv (without best-effort)
19
ipdv (with best-effort)
20
Future work

• Weighted Random Early Discard (WRED)
• interoperability testing (possible addition of
  new platforms)
• study of aggregation effects
• study of TCP related issues ACK colouring, TCP
  and UDP aggregations with AF
• service implementation and validation
• diffserv and MPLS
• diffserv and intserv
• policy management and bandwidth brokerage

21
To know more

• Interim report
• http//www.cnaf.infn.it/ferrari/tfng/ds/rep1-del.
  doc
• Diffserv test programme and interim test results
• http//www.cnaf.infn.it/ferrari/tfng/ds
• QoS measurement test programme
• http//www.cnaf.infn.it/ferrari/tfng/qosmon
• Activities of the TF-TANT task force
• http//www.dante.net/tf-tant
• tiziana.ferrari_at_cnaf.infn.it