components

enterprise-components

Kdb+ enterprise architectures for the management and analytics of large datasets are split into various components (building blocks). Each component strictly follows the defined design principles:

• public functions as fully documented APIs
• consistent logging
• interfaces for monitoring and process control

Data feeding

feedCsv - Generic CSV files reader and publisher

• provides fully customizable and configurable CSV file parser publishing data to tickLF and/or tickHF
• enables automatic detection of CSV input files in specified locations on local drive
• shields the system from corrupted data sets
• archives processed files (if required)

Data distribution

tickHF - Publishing and distribution of High Frequency data

• tuned for handling large volumes of data
• ensures minimal latency publishing
• maintains subscription lists
• provides data recovery facility (all records stored in binary journal)
• performs journal rollover at end-of-day

tickLF - Distribution of Low Frequency data

• designed for reference data handling
• handles inserts, upserts, updates and deletes
• allows data validation (e.g. model consistency check)
• allows data enrichment via user defined plugins (e.g. adding time column, adjustments of data content, etc.)
• provides data recovery facility (all records stored in binary journal)
• performs journal rollover at end of day
• maintains subscription lists
• provides ability for using custom plugins

Custom plugins are helpful tools when there is a need for data or model manipulation. For example, if received data does not contain a time column which is required in the data model, the user can use a custom plugin to dynamically add this missing column to the output table.

Data processing and storage

rdb - In-memory database

• allows configuration of subscription and end-of-day process
• performs auto re-subscription and auto reconnection mechanism to tickHF and tickLF

hdb - Historical database

hdbWriter - writes historical data directly into the hdb

• writing data directly to the hdb process
• support for writing to multiple different partitions
• support for data appending

eodMng - End-of-day processing

• provides mechanism for hdb data synchronization between different machines
• allows hdb housekeeping via predefined plugins (deletion, compression, conflation)
• exposes API for defining custom plugins

rproc - realtime processing

• performing calculation on live data streams
• subscribing to the realtime source data stream (i.e. tickHF) and calculates derived data stream
• rproc component alone does not provide any useful functionality - it is just a container
• rproc requires additional code called plugin which defines the logic for calculation of derived data
• rproc package consist of some sample predefined plugins (snap, mrvs, ohlc) which show how a plugin code can be defined. Those plugins are described in more detail below.

stream - Stream-based data processing

• allows on the fly data processing published via tickHF
• provides access to derived data via accessPoint
• allows storing derived data in rdb/hdb
• facilitates quick recovery after restart
• performs journal rollover at end-of-day

Data access

accessPoint - End users entry point

• performs user authentication and authorization
• facilitates connections to other q processes (e.g. rdb, hdb )
• enforces access control by defining functions that are permitted to specific users

System maintenance

yak - Process Management Tool

• manages (starts/stops/restarts) different type of processes
• manages group of processes (aligned with resolved dependency order)
• provides convenient access to process status and detailed process information
• offers robust configuration mechanism
• delivers bootstrapping for managed processes
• allows interrupting managed processes (UNIX only)
• supports custom pager/viewer for log files, standard output, and standard error
• provides top-like functionality

hk - Housekeeping

• performs system artifacts cleaning (compress and delete actions)
• integrated into Enterprise Components deployment structure (no cron dependencies for execution scheduling)

Monitor server

monitor Server monitoring tool

• allows capturing various information about system state, inter-process communication state, system events (like initialization of each component, subscription, journal replay), OS resources usage
• publishes data in tickHF protocol - can be stored in rdb and hdb for later analysis or processed in any tickHF protocol-compatible tool e.g. stream process

Testing

qtest - Test framework

• execution of tests organized into test suits
• test cases defined as q functions
• set of convenient assertions
• integration with enterprise-components - process management helper functions, remote assertions
• facilities for test debugging
• result export to xml file (compatible with JUnit format)

mock - Set of mock used for various tests

• mock/gen - dummy data-stream generator
• mock/empty - empty process with configuration loaded
• mock/hdbMock - hdb mock
• mock/tickHFMock - tickHF mock
• mock/rdbPluginMock - rdb plugin mock