Go

GreptimeDB offers ingester libraries for high-throughput data writing. It utilizes the gRPC protocol, which supports schemaless writing and eliminates the need to create tables before writing data. For more information, refer to Automatic Schema Generation.

The Go ingester SDK provided by GreptimeDB is a lightweight, concurrent-safe library that is easy to use with the metric struct.

Quick start demos

To quickly get started, you can explore the quick start demos to understand how to use the GreptimeDB Go ingester SDK.

Installation

Use the following command to install the GreptimeDB client library for Go:

  1. go get -u github.com/GreptimeTeam/[email protected]

Import the library in your code:

  1. import (
  2. greptime "github.com/GreptimeTeam/greptimedb-ingester-go"
  3. "github.com/GreptimeTeam/greptimedb-ingester-go/table"
  4. "github.com/GreptimeTeam/greptimedb-ingester-go/table/types"
  5. )

Connect to database

If you have set the --user-provider configuration when starting GreptimeDB, you will need to provide a username and password to connect to GreptimeDB. The following example shows how to set the username and password when using the library to connect to GreptimeDB.

  1. cfg := greptime.NewConfig("127.0.0.1").
  2. // change the database name to your database name
  3. WithDatabase("public").
  4. // Default port 4001
  5. // WithPort(4001).
  6. // Enable secure connection if your server is secured by TLS
  7. // WithInsecure(false).
  8. // set authentication information
  9. // If the database doesn't require authentication, just remove the WithAuth method
  10. WithAuth("username", "password")
  11. cli, _ := greptime.NewClient(cfg)

Data model

Each row item in a table consists of three types of columns: Tag, Timestamp, and Field. For more information, see Data Model. The types of column values could be String, Float, Int, Timestamp, etc. For more information, see Data Types.

Low-level API

The GreptimeDB low-level API provides a straightforward method to write data to GreptimeDB by adding rows to the table object with a predefined schema.

Create row objects

This following code snippet begins by constructing a table named cpu_metric, which includes columns host, cpu_user, cpu_sys, and ts. Subsequently, it inserts a single row into the table.

The table consists of three types of columns:

  • Tag: The host column, with values of type String.
  • Field: The cpu_user and cpu_sys columns, with values of type Float.
  • Timestamp: The ts column, with values of type Timestamp.
  1. // Construct the table schema for CPU metrics
  2. cpuMetric, err := table.New("cpu_metric")
  3. if err != nil {
  4. // Handle error appropriately
  5. }
  6. // Add a 'Tag' column for host identifiers
  7. cpuMetric.AddTagColumn("host", types.STRING)
  8. // Add a 'Timestamp' column for recording the time of data collection
  9. cpuMetric.AddTimestampColumn("ts", types.TIMESTAMP_MILLISECOND)
  10. // Add 'Field' columns for user and system CPU usage measurements
  11. cpuMetric.AddFieldColumn("cpu_user", types.FLOAT)
  12. cpuMetric.AddFieldColumn("cpu_sys", types.FLOAT)
  13. // Insert example data
  14. // NOTE: The arguments must be in the same order as the columns in the defined schema: host, ts, cpu_user, cpu_sys
  15. err = cpuMetric.AddRow("127.0.0.1", time.Now(), 0.1, 0.12)
  16. err = cpuMetric.AddRow("127.0.0.1", time.Now(), 0.11, 0.13)
  17. if err != nil {
  18. // Handle error appropriately
  19. }

To improve the efficiency of writing data, you can create multiple rows at once to write to GreptimeDB.

  1. cpuMetric, err := table.New("cpu_metric")
  2. if err != nil {
  3. // Handle error appropriately
  4. }
  5. cpuMetric.AddTagColumn("host", types.STRING)
  6. cpuMetric.AddTimestampColumn("ts", types.TIMESTAMP_MILLISECOND)
  7. cpuMetric.AddFieldColumn("cpu_user", types.FLOAT)
  8. cpuMetric.AddFieldColumn("cpu_sys", types.FLOAT)
  9. err = cpuMetric.AddRow("127.0.0.1", time.Now(), 0.1, 0.12)
  10. if err != nil {
  11. // Handle error appropriately
  12. }
  13. memMetric, err := table.New("mem_metric")
  14. if err != nil {
  15. // Handle error appropriately
  16. }
  17. memMetric.AddTagColumn("host", types.STRING)
  18. memMetric.AddTimestampColumn("ts", types.TIMESTAMP_MILLISECOND)
  19. memMetric.AddFieldColumn("mem_usage", types.FLOAT)
  20. err = memMetric.AddRow("127.0.0.1", time.Now(), 112)
  21. if err != nil {
  22. // Handle error appropriately
  23. }

Insert data

The following example shows how to insert rows to tables in GreptimeDB.

  1. resp, err := cli.Write(context.Background(), cpuMetric, memMetric)
  2. if err != nil {
  3. // Handle error appropriately
  4. }
  5. log.Printf("affected rows: %d\n", resp.GetAffectedRows().GetValue())

Streaming insert

Streaming insert is useful when you want to insert a large amount of data such as importing historical data.

  1. err := cli.StreamWrite(context.Background(), cpuMetric, memMetric)
  2. if err != nil {
  3. // Handle error appropriately
  4. }

Close the stream writing after all data has been written. In general, you do not need to close the stream writing when continuously writing data.

  1. affected, err := cli.CloseStream(ctx)

High-level API

The high-level API uses an ORM style object to write data to GreptimeDB. It allows you to create, insert, and update data in a more object-oriented way, providing developers with a friendlier experience. However, it is not as efficient as the low-level API. This is because the ORM style object may consume more resources and time when converting the objects.

Create row objects

  1. type CpuMetric struct {
  2. Host string `greptime:"tag;column:host;type:string"`
  3. CpuUser float64 `greptime:"field;column:cpu_user;type:float64"`
  4. CpuSys float64 `greptime:"field;column:cpu_sys;type:float64"`
  5. Ts time.Time `greptime:"timestamp;column:ts;type:timestamp;precision:millisecond"`
  6. }
  7. func (CpuMetric) TableName() string {
  8. return "cpu_metric"
  9. }
  10. cpuMetrics := []CpuMetric{
  11. {
  12. Host: "127.0.0.1",
  13. CpuUser: 0.10,
  14. CpuSys: 0.12,
  15. Ts: time.Now(),
  16. }
  17. }

Insert data

  1. resp, err := cli.WriteObject(context.Background(), cpuMetrics)
  2. log.Printf("affected rows: %d\n", resp.GetAffectedRows().GetValue())

Streaming insert

Streaming insert is useful when you want to insert a large amount of data such as importing historical data.

  1. err := cli.StreamWriteObject(context.Background(), cpuMetrics)

Close the stream writing after all data has been written. In general, you do not need to close the stream writing when continuously writing data.

  1. affected, err := cli.CloseStream(ctx)

Ingester library reference