Axibase Time Series Database

Axibase Time Series Database represents a new class of distributed systems designed to store and analyze time series data at scale.

Unlike relational databases, ATSD provides extensive APIs to simplify the process of consolidating timestamped data from multiple sources in a single repository.

Once you have statistics in one place, ATSD can help you build analytics and monitoring applications
by providing built-in Visualization for dashboards, integrated Rule Engine for alerting and automation, and Forecasting for operational analytics.

Streaming Data

Stream high-frequency data into ATSD via TCP, UDP, and HTTP protocols using text, JSON, nmon, and CSV formats.

$echo series e:srv-1 m:cpu_b=12.2 > /dev/tcp/atsd/8081
$echo "csv p:iso-parser" | cat - data.csv > /dev/tcp/atsd/8081

Batch Data

Upload CSV files directly into the database for bulk import. Use network API to upload nmon archives with wget/curl or telnet.

Query data on schedule from web services, FTP/SFTP/SCP servers, and network devices using industry-standard protocols : JMX, SNMP, CSV/TSV, and JSON.

Offload and historize data from relational databases.

Rule Engine

Create alerting rules in the built-in rule engine with time/count-based sliding windows, aggregation, and forecasting functions. Deliver alerts to enterprise consoles, email, ticketing systems, or execute system commands.

Rich Schema

ATSD is not limited to numeric data. You can store configuration files, properties, and log messages in the database as well.

ATSD supports tags everywhere – you can tag entities, entity groups, metrics, and data itself for correlation in the rule-engine and analytical queries.

List of aggregators includes COUNTER, DELTA, MAX/MIN_VALUE_TIME, and THRESHOLD functions with interpolation and windowing options.

Extensive API

ATSD Data API and Meta API implement RESTful API methods allowing you to create, edit, update, and delete meta-data such as device properties and metric settings, as well as to insert and query series, properties, and messages.

Query Language

SQL for scheduled reporting and interactive analysis by data scientists and report developers.

SELECT entity, datetime, AVG(cpu_busy.VALUE), 
   MAX(disk_used.VALUE), tags.*
   FROM cpu_busy
   JOIN USING entity disk_used
WHERE datetime BETWEEN now - 1*DAY AND now AND entity LIKE 'NURSWG*'
   GROUP BY entity, tags, PERIOD(1 HOUR)
   HAVING AVG(cpu_busy.VALUE) > 20
ORDER BY entity, datetime
   LIMIT 100

Non-parametric Forecasting

Built-in Holt-Winters and ARIMA forecasts allow you to quickly compute expected system state and make proactive decisions if observed values are outside of the confidence interval. In autopilot mode, ATSD is capable of identifying the best parameters for each algorithm, which greatly improves forecast accuracy.