YATsDb - Yet Another TimeSeries Database in .NET

Yet Another TimeSeries Database is experimental time series database inspired InfluxDb and RavenDB.

YATsDb is implement in pure .NET and modern C# with great performance, (around 30K inserts per secund on Intel i7-8550U CPU, 1.80GHz, 24GB RAM, SSD - iserting rows one at a time).

I needed a timeseries database for home IoT projects, but using InfluxDb would be boring. The implementation of the prototype, which knew how to enter data and make inquiries, took only 12 hours.

YATsDb Features

• TimeSeries database engine (indexing using tags and time, querying)
• Javascript jobs schedule using cron expressions
• Management GUI in Blazor server
• Open API definition

YATsDb is a complete database implementation with GUI, OpenAPI definitions and jobs. It requires .NET 8 runtime to run.

YATsDb.Lite Features

• TimeSeries database engine (indexing using tags and time, querying)
• AOT compiled

YATsDb.Lite is AOT compiled version with core time series database functionality. It has no dependencies on target systems.
The resulting binary is relatively small < 15MB.

Use with cURL

The following is an example of using YATsDb/YATsDb.Litle using cURL (YATsDb also provides an OpenAPI endpoint definition).

In the examples, the base URL is http://localhost:8080/.

Create bucket

Create a new bucket smart_home:

curl -XPOST http://localhost:8080/management/bucket --header "Content-Type: application/json" --data-binary "{\"bucketName\":\"smart_home\",\"description\": null}"

Insert data

Enter sample data in simple line format (<measurement>[,<tag>] <value>[,<value>...n] [<timestampInMs>]), this format is inspired by InfluxDb, but does not use named values, but positional values. Also, before InfluxDb, the unix timestamp in milliseconds is always used as the time.

The following sample data represent data from home measurement of sensors - temperature, humidity, light intensity.

curl -XPOST http://localhost:8080/write/smart_home --data-binary "sensors_data,Room1 22.0,45,80.4 1719242779202"
curl -XPOST http://localhost:8080/write/smart_home --data-binary "sensors_data,Room1 23.1,39.4,75.0 1719243079200"
curl -XPOST http://localhost:8080/write/smart_home --data-binary "sensors_data,Room1 23.2,41.7,NULL 1719243379246"
curl -XPOST http://localhost:8080/write/smart_home --data-binary "sensors_data,Room2 23.0,41,80.4 1719243379256"
curl -XPOST http://localhost:8080/write/smart_home --data-binary "sensors_data,Room1 22.0,45,80.4 1719243679194"
curl -XPOST http://localhost:8080/write/smart_home --data-binary "sensors_data,Room2 22.7,47.4,NULL 1719243979257"
curl -XPOST http://localhost:8080/write/smart_home --data-binary "sensors_data,Room1 22.1,47,NULL 1719244279035"
curl -XPOST http://localhost:8080/write/smart_home --data-binary "sensors_data,Room3 18.7,NULL,NULL 1719244579031"
curl -XPOST http://localhost:8080/write/smart_home --data-binary "sensors_data,Room1 24.7,51.7,2.4 1719244879109"

It is possible to send multiple lines in one message, but not every shell supports this.

curl -XPOST http://localhost:8080/write/smart_home --data-binary "sensors_data,Room1 24.7,51.7,2.4 1719244879109
sensors_data,Room1 24.8,55.7,3.9 1719244899109"

Query data

YATsDb allows querying in a syntax that resembles SQL, but uses value positions instead of names. The result of the query is a JSON result.

Select humidity data from Room2 older than 12 hours:

curl -XPOST http://localhost:8080/query/raw/smart_home --data-binary "SELECT AVG(1), MIN(1), MAX(1), COUNT(1) FROM sensors_data WHERE TAG('Room2') AND INTERVAL(NULL, -10h) GROUP BY +10m"

Simple line protocol

Complete grammar for inserting data data - simple line protocol:

<lines> ::= <line> [ [\r]\n...n ]
<line> ::= <measurement> <values> [<timestamp>]
<measurement> ::= measurement_name [, tag]
<values> ::= <value> [, ...250]
<value> ::={ double_value | NULL }
<timestamp> ::= unix timestamp in milliseconds

Examples

Measurement without tag:

sensors_data 12.5 1608852317000

Measurement with tag:

sensors_data,home/room3/esp32 12.5 1608852317000

Measurement with multiple values:

sensors_data,home/room3/esp32 12.5,-14.3,89,47.0036 1608852317000

Measurement with missing values (use NULL for missing value):

sensors_data,home/room4/esp32 12.5,NULL,89,47.0036 1608852317000

Measurement without unix timestamp - the current server time is used:

sensors_data,home/room3/esp32 12.5

Multiline values:

sensors_data,home/room3/esp32 12.5 1608852317000
pc_data,Workstation1 25.78,47.5,236566 1608852317900
sensors_data,home/room3/esp32 13.5 1608852417000

Select statements

Complete grammar for querying data:

<SELECT statement> ::= SELECT <projection statement> FROM measurement_name 
    [<WHERE statement>]
    [<GROUPBY statement>]
    [<LIMIT statement>]
<projection statement> ::= { * | <projection> [ ,...250 ]}
<projection> ::= <projection function>(<index>)
<projection function> ::= { VARIANCE | SUM | COUNT | AVG | IDENTITY | MAX | MIN | STDDEV | REMEDIAN | SIGN | CHANGE_PER_SEC | CHANGE_PER_HOUR | CHANGE_PER_DAY }
<index> ::= integer_number
<WHERE statement> ::= WHERE <where function> [AND ...n]
<where function> ::= { TAG(<string>) | INTERVAL(<interval exp>, <interval exp>) }
<string> ::= string
<interval exp> ::= { integer_number | <time span> | <string> | NOW() }
<time span> ::= { + | - }double_value{ ms | s | m | h | d | w | y }
<GROUPBY statement> ::= GROUP BY <time span>
<LIMIT statement> ::= LIMIT { integer_number | integer_number, integer_number | ..integer_number }

Examples

Select all data:

SELECT * FROM sensors_data

Select first 5 record from Room3:

SELECT * FROM sensors_data WHERE TAG('Room3') LIMIT 0,5

Select temperature data group by 10 minutes:

SELECT AVG(0), MIN(0), MAX(0) FROM sensors_data GROUP BY +10m

Select humidity data from Room2:

SELECT AVG(1), MIN(1), MAX(1), COUNT(1) FROM sensors_data WHERE TAG('Room2') GROUP BY +1d

Select humidity data from Room2 older than 12 hours:

SELECT AVG(1), MIN(1), MAX(1), COUNT(1) FROM sensors_data WHERE TAG('Room2') AND INTERVAL(NULL, -12h) GROUP BY +10m

Select last record:

SELECT * FROM sensors_data LIMIT ..1

Group all data:

SELECT AVG(0), AVG(1), AVG(2) FROM sensors_data

Select by concrete time (using unix timestamp in milliseconds):

SELECT AVG(0), COUNT(0) FROM sensors_data WHERE INTERVAL(1603519860000, 1608852317000) GROUP BY +10m

Select by concrete time (using human readable time):

SELECT AVG(0), COUNT(0) FROM sensors_data WHERE INTERVAL('2012-06-12 08:33:12.478', '2012-06-13') GROUP BY +10m

Potential roadmap/future work

• Charts in YATsDb
• Authentication (eg. API key, HMAC)
• UDP endpoint
• Geo-indexing

Known Issues

• Problems updating a javascript job.

Links

Technology

• ZoneTree - Thanks @koculu for the valuable advices.
• Jint
• Piglet

Inspiration

• InfluxDb
• RavenDb