Key Features

Build, deploy and scale analytic use cases fast

Modern use case development platform

Joule has been developed to leverage mature and emerging technologies through the application of clean interfaces via the Joule SDK.

Low Code

At its core, Joule adheres to the design principle of delivering a low-code use case platform that fosters rapid development iterations for impactful business outcomes. Packaged with a dedicated use case language, DSL, and a suite of reusable assets, Joule empowers developers to commence building immediately after installation.

Key language elements

Flexible event subscriptions and publishing
Stream event processing pipeline
Custom SQL Metrics definition
Extendability through custom components
Mainstream product integrations

Example

stream:
  name: standardQuoteAnalyticsStream
  enabled: true
  validFrom: 2020-01-01
  validTo: 2025-12-31
  eventTimeType: EVENT_TIME
  sources: [ nasdaq_quotes_stream ]

  initialisation:
    sql import:
      schema: standardQuoteAnalyticsStream
      parquet:
        -
          table: quote
          asView: false
          files: ['quotes*.parquet']
          index:
            fields: ['symbol']
            unique: false

  processing unit:
    metrics engine:
      runtime policy:
        frequency: 1
        startup delay: 2
        time unit: MINUTES

      foreach metric compute:
        metrics:
          -
            name: BidMovingAverage
            metric key: symbol
            table definition: standardQuoteAnalyticsStream.BidMovingAverage (symbol VARCHAR, avg_bid_min FLOAT, avg_bid_avg FLOAT,avg_bid_max FLOAT)
            query:
              SELECT symbol,
              MIN(bid) AS 'avg_bid_min',
              AVG(bid) AS 'avg_bid_avg',
              MAX(bid) AS 'avg_bid_max'
              FROM standardQuoteAnalyticsStream.quote
              WHERE
              ingestTime >= date_trunc('minutes',now() - INTERVAL 2 MINUTES) AND ingestTime <= date_trunc('minutes',now())
              GROUP BY symbol
              ORDER BY 1;
            truncate on start: true
            compaction policy:
              frequency: 8
              time unit: HOURS

    pipeline:
      - filter:
          expression: "symbol != 'A'"

  emit:
    select: "symbol, BidMovingAverage.avg_bid_max;WHERE symbol=${symbol} 'avg_bid_max'"

  group by:
    - symbol

Stream Processing

Event processing is executed through the definition of a processor's pipeline. Events undergo sequential processing utilising a micro-batch methodology, a technique employed to boost processing throughput while optimising the utilisation of underlying hardware capabilities.

Filtering

Filtering event based using a configurable criteria or a an expression. Example use cases is customer opt-out, missing data elements, out of range etc,.

Enrichment

Enrichment of streaming events with a embedded low-latency data caching solution

Transformation

Event field tokenisation, encryption, masking, bucketing and redaction

Triggers

Real-time alerts and event triggers using rule based processing and delta CDC processing

Event Tap

Tap events directly in to an in-memory database to enable on/off line processing

Scripting

Execute external scripts or defined expression within the use case DSL. using supported scripting languages ( Node.js, JavaScript, Python)

Metrics

A SQL compliant metrics engine which computes scheduled metrics

Machine Learning

Leverage streaming online predictions to drive insights to action

Analytics

Streaming analytics using event windows, expressions, scripts and much more

New processors are constantly added to the platform. Please contact fractalworks for an updated list.

Analytics

Joule provides three flexible methods to build analytical insights. Each method is describe below.

The integration of streaming analytics serves as a pivotal feature, empowering the evolution of sophisticated use case development, including applications like geospatial analytics for marketing, business analytics, and feature preparation for machine learning predictions.

Key Features

Tumbling and sliding windows
Standard statistical functions
Custom analytic functions
Geospatial analytics (Geo Tracker, Geofence occupancy, spatial index)

Example

time window:
  emitting type: tumblingQuoteAnalytics
  aggregate functions:
    FIRST: [ask]
    LAST: [ ask ]
    MIN: [ ask ]
    MAX: [ bid ]
    SUM: [ volume ]
    MEAN: [ volatility ]
    VARIANCE: [ volatility ]
    STDEV: [ volatility ]
  policy:
    type: tumblingTime
    window size: 5000

Utilising machine learning models enables the delivery of streaming predictions, offering versatility for applications in both start-of-day batch processing and real-time prediction services to bolster intra-day business operations.

Key Features

Predictive Model Markup Language (PMML)
In-place model refresh / replacement
Model versioning control
Prediction and model auditing
Feature engineering (binning, scaling, log transform, and scripting)

Example

pmml predictor:
  name: irisScorer
  model: ./models/iris_rf.pmml
  response field: flowerPrediction
  
  audit configuration:
    target schema: ml_audit
    queue capacity: 5000
    flush frequency: 300

Joule offers a comprehensive global metrics engine designed to facilitate the development of intricate use cases. Developers can articulate metrics using standard ANSI SQL and specify their execution frequency using the Joule DSL. These defined metrics are seamlessly integrable into the defined streaming pipeline.

Key Features

ANSI SQL based metrics
Runtime policy to schedule metric generation
Metrics accessible within pipeline processes and projection
Export RestAPI
In-memory column database

Example

metrics engine:
    runtime policy:
      frequency: 1
      startup delay: 2
      time unit: MINUTES

    foreach metric compute:
      metrics:
        -
          name: BidMovingAverage
          metric key: symbol
          table definition: standardQuoteAnalyticsStream.BidMovingAverage (symbol VARCHAR, avg_bid_min FLOAT, avg_bid_avg FLOAT,avg_bid_max FLOAT)
          query:
            SELECT symbol,
            MIN(bid) AS 'avg_bid_min',
            AVG(bid) AS 'avg_bid_avg',
            MAX(bid) AS 'avg_bid_max'
            FROM standardQuoteAnalyticsStream.quote
            WHERE
            ingestTime >= date_trunc('minutes',now() - INTERVAL 2 MINUTES) AND ingestTime <= date_trunc('minutes',now())
            GROUP BY symbol
            ORDER BY 1;
          truncate on start: true
          compaction policy:
            frequency: 8
            time unit: HOURS

Observability

Each processing component in Joule furnishes a standard set of metrics, offering users insights into he number of events received, processed, discarded, and failed. Furthermore, with the SQL engine enabled, both raw and processed events are stored, making them queryable and exportable for enhanced analytical capabilities.

Integrations

Straight out of the box, Joule provides a standardised set of data integrations, facilitating a quick start to data consumption and publication processes.

Flexible Deployment

Joule has been designed to be platform agnostic, offering seamless deployment options whether you choose a local, on-premise, or cloud-based environment. Joule is packaged as a Docker container for simplified deployment configurations or as a standalone binary, providing flexibility to meet diverse deployment needs.

SDK

A Java SDK for developers is supplied to extend platform capabilities, enabling the customisation and enhancement of processors and data transports.

PreviousOverview NextFAQ

Last updated 3 months ago