CWIC Metrics Eugene G. Yu ([email protected])[email protected] Lingjun Kang...

8
CWIC Metrics Eugene G. Yu ([email protected] ) Lingjun Kang ( [email protected] ) Archie Warnock ([email protected] )

Transcript of CWIC Metrics Eugene G. Yu ([email protected])[email protected] Lingjun Kang...

Page 2: CWIC Metrics Eugene G. Yu (gyu@gmu.edu)gyu@gmu.edu Lingjun Kang (lkang3@masonlive.gmu.edu)lkang3@masonlive.gmu.edu Archie Warnock (warnock@awcubed.com)warnock@awcubed.com.

Outline

• CWIC Metrics– Current architecture– Current status

• CWIC Metrics Development Plans– New capabilities– Possible extension to meet the new capabilities

Page 3: CWIC Metrics Eugene G. Yu (gyu@gmu.edu)gyu@gmu.edu Lingjun Kang (lkang3@masonlive.gmu.edu)lkang3@masonlive.gmu.edu Archie Warnock (warnock@awcubed.com)warnock@awcubed.com.

CWIC Metrics

• URLs– CSW - http://cwic.wgiss.ceos.org/cwicmetrics/– OpenSearch - http://cwic.wgiss.ceos.org/cwicosmetrics/

• What are recorded?– Dataset– Connector

• What are reported?– By dataset– By pre-defined connector (new connector needs to be added

by changing the code and re-calculation)– Reports are dynamic and interactive

Page 4: CWIC Metrics Eugene G. Yu (gyu@gmu.edu)gyu@gmu.edu Lingjun Kang (lkang3@masonlive.gmu.edu)lkang3@masonlive.gmu.edu Archie Warnock (warnock@awcubed.com)warnock@awcubed.com.

CWIC Metrics - Architecture• Logging appended into a single file• Recorded only dataset and connector• File-based summary

– Directly added to the json file– Revisit the record of the same period would be resulted in double counting

• Modifications would lead to recalculation from the very beginning

• Connectors are pre-set to be summarized– Hard-code the list of connectors to be summarized

Page 5: CWIC Metrics Eugene G. Yu (gyu@gmu.edu)gyu@gmu.edu Lingjun Kang (lkang3@masonlive.gmu.edu)lkang3@masonlive.gmu.edu Archie Warnock (warnock@awcubed.com)warnock@awcubed.com.

CWIC Metrics - status

• Dataset and their beginning dates– List of current datasets

(to be summarized and listed)

• What can be seen?– A screen capture

showing the reports• By dataset• By connector

Page 6: CWIC Metrics Eugene G. Yu (gyu@gmu.edu)gyu@gmu.edu Lingjun Kang (lkang3@masonlive.gmu.edu)lkang3@masonlive.gmu.edu Archie Warnock (warnock@awcubed.com)warnock@awcubed.com.

CWIC Metrics – new capabilities (1/2)

• New capabilities– Finer details for metrics collection

• Basic unit up to dataset• Added details of information

– Data originator– Data distributor– Projects

– Bases for metric summary• dataset (basic unit – lowest level of metric data collection)• Data centers (Distributor? Originator?)• Projects

Page 7: CWIC Metrics Eugene G. Yu (gyu@gmu.edu)gyu@gmu.edu Lingjun Kang (lkang3@masonlive.gmu.edu)lkang3@masonlive.gmu.edu Archie Warnock (warnock@awcubed.com)warnock@awcubed.com.

CWIC Metrics – new capabilities (2/2)

• New capabilities (continued)– Reflecting hierarchical ownership• Data

– Aggregation (hierarchical structures)• Aggregate by connector• Aggregate by data centers• Aggregate by project

– Any other aggregation option?

Page 8: CWIC Metrics Eugene G. Yu (gyu@gmu.edu)gyu@gmu.edu Lingjun Kang (lkang3@masonlive.gmu.edu)lkang3@masonlive.gmu.edu Archie Warnock (warnock@awcubed.com)warnock@awcubed.com.

CWIC Metrics – Architecture (updating)• To support the new capabilities• Necessary changes

– Restructure the metrics logging• Adding required info

– Finer details– Terminology alignment

• Roll-over by dates (reducing the file size)

– Collected base metrics management• DBMS or file-based database (e.g. MongoDB)

– Aggregation implementation• SQL-based aggregation (e.g. Views)• Output JSON for presentation (e.g. modular JSON converter from Views or Web Services to populate

the JavaScript module)• Combined CSW/OpenSearch metrics