COSMOSDataCollectionMeeting20061212

Agenda

• introductions
• discuss meeting schedule and agenda
• component objectives and long term roadmap
• near term deliverables

Minutes

Attendees

• Marius Slavescu (IBM)
• Sheldon Lee-Loy (IBM)
• Mark Weitzel (IBM)
• Craig Thomas (Groundwork opensource)
• Don Ebright (Compuware)

Discussion

The group introduced themselves and described their interests in the DataCollection component:

Don: Reusable data collection framework suitable for reuse in commercial products

Mark: Standards-based data collection and storage framework to standardize infrastructure across multiple products

Craig: Understanding the relationship between the COSMOS reporting and data collection components

Marius: A common solution between TPTP and COSMOS

Sheldon: Alignment between COSMOS data reporting component and TPTP

The group agreed to the proposed agenda.

We discussed the project vision and long term goals. Some of the items that were discussed are:

COSMOS needs a base set of data collection and reporting functionality that is easy to use out of the box.

The data collection framework should be reusable in commercial products.

COSMOS needs a standardized set of reports based on a pluggable data store.

A few questions were posed and the group expressed their opinions regarding the answers.

Q. What should be stored and how?

A. We should align the functional capabilities of the Data Collection component with the requirements of the Management Enablement and Data Visualization components.

Q. Which RDBMS should we use?

A. Suggestions included Derby and MySQL. TPTP is committed to delivering a persistence framework based on Derby. The goal is to minimize dependencies on a specific RDBMS.

Q. Which data collectors are required?

A. The data collection capabilities must be aligned with the requirements of the Management Enablement component. Initially, we want to reuse TPTP data sources but the framework should facilitate support of a wide variety of data sources.

Q. How is SML related?

A. The interface into the data collection will be a WS-Notification-like publication and subscription of events. These events will be in WEF (WSDM Event Format) and the extended data element will contain an SML-IF document. This will evolve over time as SML becomes more mature.

Craig presented the COSMOS end to end use case. For more information see: End-to-End_use_cases

Short term deliverables discussion:

TPTP is building an RDBMS based data collection system. COSMOS will reuse its API and as much of the code as possible.

Marius is working on using Derby for persistence of TPTP trace, log, and statistical models. The TPTP 4.4 code line is now open in CVS and Marius will be building an initial implementation for M1. The page describing the TPTP models and their persistence is here TPTPModel.

TPTP will build WEF support and COSMOS will support SML-IF as an extended data element in these events.

We will continue to discuss how to deliver a working end to end use case some time in January.

All attendees expect to be able to attend next week at the same time so the next meeting is scheduled for Tuesday December 19.

Actions

• Mark W. to work to revive the End-to-End Use Case working group
• Mark W. to establish a weekly COSMOS architecture meeting starting in January
• Mark W. to announce TPTP / COSMOS architecture alignment meeting

Misc

Mark noted that the Aperi project would like to engage with TPTP data collection when we are ready.