The DRC Novara search engine delivers inline, real-time identification of key, imprecise phrases and bit patterns using a fuzzy logic analyzer that scales to instantaneous analysis of millions of messages and datastreams without the need to index first. Enabling actionable intelligence in sub-seconds.

Background

With over 20 trillion SMS messages sent in 2015 alone (not counting Twitter messages) people are rapidly shifting P2P communications from conventional semi-structured mechanisms, like email, to unstructured, casual messaging. With social messaging shorthand, mixed languages and misspellings are common, and therefore conventional analyzers, like regular expressions, looking for exact pattern matches no longer work.

Additionally inline analysis of network and system log data frequently requires identification of imprecise or truncated strings and numbers.

Solution

The DRC Novara analyzer solves this dilemma by employing an advanced bit pattern analysis technique that recognizes imprecise matches and scores the match. This enables the user to precisely tune search criteria for the terms and expressions required.


Dense packaging.

Each Novara server contains up to four analyzer engines packaged in a 1U rack mountable configuration.

Highly flexible expression engine.

Each analyzer engine can be individually configured to handle hundreds of search strings of variable lengths simultaneously.

Inline, very low latency.

By analyzing inline incoming data in real-time, microsecond response times are achieved.

Highly tunable, multi-threshold capacity.

Various scoring thresholds can be configured enabling the user to determine when to receive a match alert.

Massive scalability.

Clustered Novara analyzers enable ten of thousands of expressions, petabytes of data and thousands of users to be supported simultaneously.

Ultra-low energy consumption.

Each Novara server with 4 analyzer engines requires less than 400 watts of power.

Cloud ready.

Novara analyzers can be cloud based.


Specification (per analyzer engine):

Number of search strings

2 to 1,000

String length (bytes)

2 to 2,048

Bytes/character

1 to 4

Penalty score range

-32 to +31

Data throughput(standard)

100 MB/s

Data throughput(custom)

2.5 GB/s

Source data size

Unlimited

1U server contains up to 4 analyzer engines.

1Us can be clustered.