Metric Composition Framework (MCF)

From GEANT2-JRA1 Wiki

1 Metric Composition Framework

Metric Composition Framework

Technical Annex GN2-JRA1

"... Work Item 2: Infrastructure Design

The work on the General Framework Design has provided the guidelines for the overall JRA1 design. During Year2, the work on detailed design of the basic services has greatly advanced. The work will mostly focus on the AA design and its integration within the existing services. This work will be done in partnership with JRA5 for the Authentication part. During Year2, the activity has produced a deliverable DJ1.2.3 describing how to concatenate and aggregate different types of metrics together. The theoretical work done so far need to be enhanced with pragmatical guidelines for the NRENs. In addition the concatenation and aggregation work will also cover sub-layer3 metrics. The outcome will be described in deliverable DJ1.2.5. Over Year2, we have discovered that we are facing a huge amount of data not manageable manually. Some work need to be done to find way of summarizing the data or extracting abnormal behaviors. The outcome of this work will be presented in deliverable DJ1.2.4.

Deliverables

DJ1.2.4 - PerfSONAR AA Service Specification (M29) ...

DJ1.2.5 Metric Composition Framework - experimental results and sub Layer3 (M33) This Deliverable intend to provide a pragmatical approach of the MCF to the NRENs and ways of summarising the information and report on the experiment done. It will also provide an extention to the MCF for selected L1 and L2 metrics.

This Work Item will be lead by GRnet with participation from Belnet, DANTE, DFN, RedIRIS and Renater. ..."

See also:

Media:MCF Y3 WBS.jpg - Indicative Work Breakdown Structure

[edit]

L1/L2 Monitoring

Provided that monitoring information for layers L1/L2 will be available by the JRA3 and JRA4 activities, the JRA1 activity has to concentrate on post-processing analysis of the available information. Data analysis can be categorized as follows:

1. Concatenation and/or aggregation of performance metrics for a single L1/L2 technology for each point-to-point service identified in GN2, a.k.a. Gigabit Ethernet services, 10 Gigabit Ethernet services, and SDH/SONET services.

2. Concatenation and/or aggregation of performance metrics for different L1/L2 technologies having implementation similarities. For example, point-to-point Gigabit Ethernet services over GN2 network terminating at a GigE interface at NREN A and SDH STM-64 interface at NREN B have to be monitored. In this case, Ethernet and SDH/SONET technologies are combined for the provision of a point-to-point service.

3. Correlation of performance metrics between L1/L2 layers and L3 layer for services over the same physical path. In this case, for example, JRA1 may try to associate error-related metrics, such as bit-error-rate (BER) at the optical transmission systems or errored seconds at the SDH/SONET systems, with the packet losses measured at layer.

Media:ICNS07 -2.pdf - Submitted to ICNS'07 - Media:Presentation.ppt - Draft presentation
Media:GN2 services & JRA1 PMs - 20061130.doc - JRA1/MCF Internal Document: Concatenation of L1/L2 metrics
Media:NEP-05-069.doc

See also:

- Lightpath Monitoring

[edit]

CNM Dashboard

The main purpose of CNM Dashboard is to provide a friendly graphical user interface (GUI) that presents in a concise way an overview of the most important performance metrics of a network . The CNM Dashboard GUI will be used for trouble identification as well as for network trend observations.

Media:CNM Dashboard V1.2.doc - Proposal document

See also:

- Key Performance Indicators

[edit]

L3 Metric Aggregation / Concatenation

The L3 Metric Aggregation / Concatenation activity aims to continue the work done in GN2-GRA1 Y2 and documented in deliverable DJ1.2.3. Network performance raw data, which will be collected under various congestion network conditions, will be analysed using the methodology presented in deliverable DJ1.2.3. Special focus will be given in error propagation.

This activity intends also to provide guidelines developers on how to fine tune the measuring infrastructure in GN2. Hades (IPPM), BWCTL, and SSH/SNMP tools are planned to be used in the tests.

Following this activity, alternative tools for Available Throughput measurements were investigated (as BWCTL is very intrusive). Pathload is found an interesting tool, an overview can be found here.

[edit]