Corvil Streams

Corvil Streams identifies key analytics for all applications from the torrent of network data.

Corvil Streams identifies key analytics for all applications from the torrent of network data — producing streams of actionable events for review in the Corvil UI or integration with external systems.

The network environment is shared by hundreds of applications with different characteristics and requirements, different users, and different problems. From the continuous flow of this application data on your network, Corvil Streams pull out the key events for each application — and generates a concise summary for each one. These events can be reviewed, filtered, and categorized in the Corvil UI or exported in real-time to third-party systems such as Splunk, Hadoop, AWS Kenesis, Tableau and STORM.

Benefits

  • Pulls out and summarizes key events for each application

  • Reformats the encoded information in an Analytics Stream into a human-readable format

  • An open and flexible API

  • Normalize large amounts of network data for consumption by northbound systems

  • Detect anomalous or erroneous conditions in network and application data — rejects, transaction failures, excessive latencies

The analytics stream for an application will contain a variety of event data: performance and health reporting will generate events for server outages and response time issues; business-level analytics will be reported through per-transaction summaries driven by the decoding of the application payload.

An analytics stream can easily be tuned — new event types can be added or existing events modified. Event fields can be populated from any of the decoded network data or the additional analytics provided by Corvil’s Plugins suite. The streams are tuned to the needs of different users — to providing the application ops team with details on sluggish response times, or the network team with updates about DNS or BGP problems.

Network data protocols exchange huge volumes of encoded information — the events in an Analytics Stream reformat into human-readable form:

  • Web Application - Report URI and payload for requests that do not meet response time targets; notification of user login and logout; record full request details when users place an order.

  • BGP - Report status of peering relationships and notify when a peered BGP router disconnects; detect route flapping and report affected IP prefixes; alert on withdrawn routes including name of affected network.

  • VoIP - A call record dashboard provides summary details: call duration and time, the caller and callee, SIP metrics relating to call set-up, MOS and jitter measured directly from the RTP stream, call quality as reported by handsets.

  • Electronic Trading - All trading activity that affects a firm’s position can be reported via a drop-copy — normalized into a standard format regardless of the underlying venue-specific binary protocol.

For high volume streams, events can also be summarized on a per-second basis to produce aggregated results. For example, the analytics stream for a web application can produce summarized results for response times and error codes on a per-URL basis.

The underlying network data that was used to produce the event is always available for drill-down when the additional context is required.

The Corvil Streams Browser

The Corvil Streams UI presents information from all streams in a clear tabulated form.

  • Events are reported live from the network and a historical record can be analyzed if required.

  • Context-sensitive drill down to the underlying network data is available for every event.

  • Quick filter options allow you to focus on streams of interest and on individual types of events within the stream.

  • Advanced filtering capabilities allow you to do targeted searches.

Integration with external systems

The analytics streams are designed for easy integration with other applications. An open and flexible API provides real-time access so that they can become an input to external dashboards, existing monitoring systems, or proprietary applications.

  • Flexible filtering capabilities are available at the API so that applications can subscribe to the appropriate substream of events.

  • Efficient encoding using the open-source Google Protocol Buffer standard.

  • Easy to integrate with sample implementations provided for Python, C++, and Java.

  • High reliability with a facility for historical replay if a client’s connection for the live stream is interrupted by the system or network problems.