Skip to content

Approximate time-series database using sliding window aggregation

License

Notifications You must be signed in to change notification settings

Squadrick/summarydb

Repository files navigation

 _____                                           ____________ 
/  ___|                                          |  _  \ ___ \
\ `--. _   _ _ __ ___  _ __ ___   __ _ _ __ _   _| | | | |_/ /
 `--. \ | | | '_ ` _ \| '_ ` _ \ / _` | '__| | | | | | | ___ \
/\__/ / |_| | | | | | | | | | | | (_| | |  | |_| | |/ /| |_/ /
\____/ \__,_|_| |_| |_|_| |_| |_|\__,_|_|   \__, |___/ \____/ 
                                             __/ |            
                                            |___/             

Go

This is an implementation of SummaryStore in Golang.

By using window sliding aggregations, SummaryDB achieves much lower disk usage and lower time-based range query latencies compared to other TSDBs. These benefits come at the cost of higher error bounds of the query results.

SummaryDB is best suited for high volumes of numerical data, and it currently allows for querying of the following metrics across time:

  1. Max
  2. Min
  3. Count
  4. Sum

On generic data, it supports:

  1. Membership (using bloom filters)

Example

package main

import (
    "context"
    "summarydb"
)

func main() {
    db := summarydb.New("/path/to/db")
    // OR
    db := summarydb.Open("/path/to/db")
    defer db.Close()

    seq := summarydb.window.ExponentialLengthsSequence(2)
    stream := db.NewStream([]string{"sum", "max"}, seq).Run()

    stream.Append(0, 10.0)
    stream.Append(1, 11.0)
    stream.Append(2, 12.0)
    stream.Append(3, 13.0)
    stream.Append(4, 14.0)

    // Get sum between t=1 and t=3
    params := QueryParams{
        ConfidenceLevel: 0.95,
        SDMultiplier:    1.0,
    }
    aggResult := stream.Query("sum", 1, 3, &params)
    result := aggResult.value.Sum.Value
    error := aggResult.error
}

Dependencies

  1. BadgerDB is the persistent key-value store.
  2. Ristretto is the cache for the backing disk storage.
  3. Cap'n Proto is the serialization format for on-disk representation for all the window data.

Languages