Contents

Introduction

Aim of Centinel is to assist operators by providing visibility into the OpenDaylight management software defined network ecosystem (network via plugins, apps and infrastructure). To accomplish this, Centinel - Boron aims at a multi phased approach that includes:

• Data Collection: IPFIX, OpenFlow (from ODL-OF Plugin) and SNMP (from ODL-SNMP Plugin)

• Data Analysis: Using Kafka, Storm/Spark, Flume (replace Graylog)

• Analytics Dashboard: Inter-op with DLUX framework and ensure UX/UI (Graphana/Kibana)

• Machine Learning: Leverage Apache Mahout/SparkMlib to enable machine learning based on above analysis

• Additionally from usage-deployment, Embedded installation is proposed (Support for Embedded flume agent. Support for embedded installation of Query engine)

• Framework enhancement -

^{Enhance Centinel architecture to have two branches A. Batch processing system(BPS)-BPS comprise of flume, hbase. B. Real time analytics system(RAS)- RAS comprise of apache strom (Embed and distributed mode) for real time prediction. In embed mode, storm will be used instead of Graylog server.In distributed mode, user will have both the options for log analysis.}

Target Use-Cases

• Bandwidth utilization across WAN link

• Performance monitoring

Release Deliverable

Release Milestones

Externally Consumable APIs

• None Listed. Centinel creates own APIs based upon relevant data required.

Expected Dependencies on Other Projects

None

Expected Incompatibilities with Other Projects

None

Compatibility with Previous Releases

Changed APIs and/or Functionality

• Not applicable. This would be first release of Centinel in OpenDaylight Boron release.

Themes and Priorities

Requests from Other Projects

None Listed

Test Tools Requirements

• Java unit and integration tests

Other

See also

• Centinel Home Page

• Centinel Project Proposal

• Centinel Boron Release Plan

• Centinel-dev Archives

• File:Centinel - Design document M2Sep072015-InitialDraft.pdf