EGEE-II INFSO-RI-031688 Enabling Grids for E-sciencE End-to-End Service Level Agreement Provisioning and Monitoring for End-to-End QoS

EGEE-II INFSO-RI-031688

Enabling Grids for E-sciencE

www.eu-egee.org

End-to-End Service Level Agreement Provisioning and Monitoring

for End-to-End QoS

TNC2007, 21-24 May 2007

Vassiliki Pouli (GRNET/NTUA) [email protected]

TNC2007, 21-24 May 2007 2



Outline

• Introduction

• SLA parts

• Model of SLA establishment

• Monitoring of SLAs

• Questions

TNC2007, 21-24 May 2007 3



Introduction (1)

• Grid applications require specific level of QoS for data transfers across different Resource Centres (RCs)

• end-to-end (e2e) SLA is needed to define this level of QoS – provides the technical and administrative details to perform

Maintenance Monitoring Troubleshooting

TNC2007, 21-24 May 2007 4



Introduction (2)

• Grid RCs: users of the network• Network providers

– GÉANT (paneuropean network)– NRENs (national networks)– Regional/metropolitan/campus networks

• Synthesis of e2e SLA based on individual domain SLAs

TNC2007, 21-24 May 2007 5



SLA parts

• ALO (Administrative Level Object) – Contacts– Duration (start and end times)– Availability of service– Response times– Fault handling procedures

• SLO (Service Level Object) – Service flow description– Excess traffic treatment– Performance guarantees (OWD, jitter, packet loss, capacity…)– Reliability guarantees: max downtime (MDT), time to repair (TTR)– Monitoring infrastructure

TNC2007, 21-24 May 2007 6



Trusted 3rd party - ENOC

• Multi-domain SLA• Need for a trusted 3rd party responsible for:

– SLA service installation– SLA management/monitoring– User support for network issues

network SLA violations

• For EGEE ►►► ENOC (EGEE Network Operation Centre)– Counterpart of a NREN NOC for EGEE– In charge of “EGEE network” operations

TNC2007, 21-24 May 2007 7



Model of SLA establishment(1/2) Preliminary agreement

1. ENOC asks from every participating domain and RC to formulate an agreement

2. Each domain NOC provides– the ALO (Administrative Level Object)– max bandwidth allocated for EGEE

Each RC – provides administrative and technical

details – signs Acceptable Use Policy (AUP)

Provisioned network resources used only for EGEE purposes

3. ENOC stores the received information to the NOD (Network Operational Database)

Preliminary agreement

•Preliminary agreement of ENOC with participating domains & RCs• made once for every participating domain & RC

TNC2007, 21-24 May 2007 8



Model of SLA establishment (2/2)2-Stage provisioning model: SR-SA

2-Stage Provisioning Model: SR-SA• Stage 1: Service Reservation (SR) – b2b SLA

– PIP (Premium IP) reservation in extended QoS network (GEANT/NRENs)– border-to-border SLA (GEANT/NRENs SLAs)

• Stage 2: Service Activation (SA) – e2e SLA– Activation of the service ↔ Configuration of the routers in the last mile

network– end-to-end SLA (b2b SLA + NREN client domains’ SLAs)

TNC2007, 21-24 May 2007 9



Rationale of 2 stage provisioning model

• Grid applications require e2e path be available in time

• Lead time between service request and service reservation of the extended QoS network– Manual configuration of the routers – Currently 2 working days in the GEANT network

TNC2007, 21-24 May 2007 10



Service Reservation – b2b SLA

• Case 1: automatic reservation– Reservation via AMPS servers of hosting NRENs and GEANT– AMPS (Advanced Multi-domain Provisioning System) :

System In development stage by the GEANT project Management of the whole PIP provisioning process from user

request through to the configuration of the appropriate network elements

• Case 2: manual reservation– No AMPS servers (or similar services) installed

TNC2007, 21-24 May 2007 11



Stage 1: Service Reservation (SR)-b2b SLAcase 1: automatic reservation

• Reservation through AMPS or similar service

• ENOC identifies involved GEANT/NREN domains

• GEANT/NRENs provide individual SLAs

• Synthesis of b2b SLA: performed by ENOC based on reported GEANT/NRENs SLAs

TNC2007, 21-24 May 2007 12



Stage 1: Service Reservation (SR)-b2b SLAcase 2: manual reservation

• Cases with no AMPS servers or similar services installed in NRENs

GEANT/

NRENs

TNC2007, 21-24 May 2007 13



Stage 1: Service Reservation (SR)-b2b SLAcase 2: manual reservation

• No AMPS servers installed• ENOC identifies involved

GEANT/NREN domains• ENOC initiates manual requests

to individual domain NOCs• NOCs reply by email and provide

individual SLAs• Synthesis of b2b SLA: performed

by ENOC based on reported domain SLAs

TNC2007, 21-24 May 2007 14



Stage 2: Service Activation (SA)-e2e SLA

• ENOC verifies that the reservation in the extended QoS domain is still effective and retrieves it

• Checks if NREN client domains (MAN/campus/institution) can support the request

• NREN client domains provide their SLAs

• ENOC produces e2e SLA based on:– reported NREN client domains’ SLAs – b2b SLA from stage 1

TNC2007, 21-24 May 2007 15



Monitoring of SLAs• ENOC queries NPM DT (Network

Performance Monitoring Diagnostic Tool)

• NPM DT provides measurement data from perfSONAR (GEANT/NRENs) and e2emonit (RC-to-RC) monitoring frameworks

• Fault Identification/Notification – Case 1: ENOC identifies & notifies

responsible domain – Case 2: ENOC (not able to isolate

the problem) informs GEANT PERT (Performance Enhancement Response Team)

• Reaction-Repair according to SLAs• ENOC checks SLA compliance

TNC2007, 21-24 May 2007 16



SLA monitoring requirements

e2e Metrics:– OWD (One Way Delay)– IPDV (IP Packet Delay Variation)– RTT (Round Trip Time)– Packet Loss– Available bandwidth– Achievable bandwidth– TTR (Time To Repair)

From trouble ticket issue to recovery, per violation

– MDT (Maximum DownTime)Maximum total TTRs for all violations in a given period

Performance metrics

Reliability metrics

TNC2007, 21-24 May 2007 17



– Frequent e2e and partial domain monitoring of performance metrics (e.g. every 15’) in agreed service availability period

– Capability of setting thresholds on metrics to generate violation alarmsDifferent severity levels (?)

– Trouble tickets, triggered by users and ENOC operators on alarms, managed via TTM (Trouble Ticket Manager)

– Statistics from trouble tickets to infer MDT & TTR

Monitoring features

TNC2007, 21-24 May 2007 18



Questions

?

Documents

EGEE-II INFSO-RI-031688 Enabling Grids for E-sciencE End-to-End Service Level Agreement Provisioning and Monitoring for End-to-End QoS