Continuum

continuum is a JVM library for storing and analyzing large amounts of time based data.

• Core library for storage, retreival, and analysis of time series and time key value data
• REST Interface
• Streaming backup/restore
• Streaming master/slave and master/master replication

tl;dr

Building

make

or

./gradlew install

Maven

		<dependency>
			<groupId>continuum</groupId>
			<artifactId>core</artifactId>
			<version>0.+</version>
		</dependency>

Javadocs available on the web

Goals

1. Use memory mapped files and log-structured-merge trees (LSM) to efficienty store, retreive, and analyze large amounts of time-based data.
2. Leverage existing underlying core technologies like RocksDB, LevelDB, BerkeleyDB and similar datastores.
3. Employ time-to-live (TTL) or downsampling (RRD) to efficeintly use disk space.
4. A continuous, scalable infintie time-based data stream by incorporating data windowing, downsampling, or both
5. Utilize TWO specialized schema designs for time series data and time key value data.

TimeSeries Data (space/time): (Analysis and Dashboard/UI)

• Small number of unique time series with very large data volume (millions to trillions)
• Infinite storage using retention policies and data downsampling (RRDTool, Whisper, InfluxDB inspired)
• Downsampling OK
• Example: "Realtime" performance metrics, measurements of cpu, or weather temperature readings over time

TimeKeyValue Data (time/key:value): (User Session and Session metrics)

• Large number of unique keys with smaller amount of data (10s to 1000s)
• For aggregating small amounts of data into buckets by a unique key
• Age out over time
• Stream
• Fast data: hot, open sets of newest data
• Slow data: rolling window of data through ColdStorage (S3, Hadoop, NAS)
• Example: "Realtime" analytics data events grouped by session or user id

Scaling

• add disks
• cold online backup (backup/restore)
• hot online backup(s) (read-only replication)
• application level sharding by key or time
• cluserting (master-master replication)

Maintainer

Zack Bartel zack@bartel.com zackbartel.com