Manage a metricstore

A metricstore collects, stores, and queries metrics in SLS. Each metricstore belongs to a project, which can contain multiple metricstores. For more information, see Metricstore.

Create a metricstore

1. Log on to the Simple Log Service console.

2. In the Projects section, click the one you want.

   

3. On the Metric Storage > Metricstores tab, click the + icon.

4. In the Create Metricstore panel, configure the following parameters and click OK.

   Parameter	Description
   Metricstore Name	The metricstore name. Must be unique within the project. Cannot be changed after creation.
   Data Retention Period	How long metrics are retained in the metricstore. Specified Days: Retains metrics for 1 to 3000 days. Warning Expired metrics are automatically deleted. If you shorten the retention period, SLS deletes expired metrics within 1 hour. The Usage Details on the SLS console homepage updates the next day. For example, changing the retention period from 5 to 1 day deletes the previous four days of metrics within 1 hour. Permanent Storage: Retains metrics indefinitely. Note A retention period of 3650 returned by the SDK indicates permanent storage.
   Shards	The number of shards. Each shard supports 5 MB/s and 500 writes/s for writing, and 10 MB/s and 100 reads/s for reading. Maximum: 10 shards per metricstore, 200 per project. For more information, see Shard.
   Automatic Sharding	If you turn on Automatic Sharding, SLS increases shards when write capacity is insufficient. For more information, see Manage shards.
   Maximum Shards	Required when Automatic Sharding is enabled. Maximum number of read/write shards. Maximum value: 256.

Modify a metricstore

1. On the Metric Storage > Metricstores tab, hover over the target metricstore and choose > Modify.

2. On the Metricstore Attribute page, click Modify.

   • Basic Information

     • Data Retention Period: For a description of the parameters, see Create a metricstore.

     • Automatic Sharding: Enable this feature to automatically split shards and increase write throughput. For more information, see Manage shards.

     • Maximum Shards: Maximum number of read/write shards for automatic splitting. Maximum: 256.

     • Log Public IP: When you enable Log Public IP, SLS adds the following fields to each log entry:

       • __client_ip__: The public IP address of the device where the log originates.

       • __receive_time__: The time when the log arrives at the server, as a UNIX timestamp (seconds since January 1, 1970, 00:00:00 UTC).

   • Shard Management

     When a metricstore is created, two shards are created by default. Split or merge shards later as needed. For more information, see Manage shards.

   • Query Acceleration Settings

     By default, the Prometheus Query engine does not cache results and runs single-coroutine calculations on a single node, which leads to poor performance with many timelines, long query time ranges, or complex queries. SLS introduces Global Caching and Concurrent Computing to improve PromQL performance. For more information about the low-level design and configuration, see Query acceleration.

   • Write Settings

     Metricstore stores metric data chronologically. Out-of-order writes (such as writing months-old data or data with incorrect timestamps due to clock drift) can severely degrade query performance.

     Configure left and right time windows (in seconds) to filter data with abnormal timestamps. The valid range is [arrival time − left interval, arrival time + right interval]. Data outside this range is discarded. Setting [0,0] disables the filter.

     Note

     This feature applies only to data written via the Prometheus Remote Write protocol. For more information about connection types, see Collect metric data from Prometheus by using the Remote Write Protocol.

   • Ingest Processor

     Processes data before writing, supporting field modification, parsing, filtering, and masking. For more information, see Process data during ingestion (ingest processor).

   • Tag

     Add tags to a metricstore to group and manage your metricstores.

3. Click Save.

Delete a metricstore

1. On the Time Series Storage > Metricstore tab, move the pointer over the metricstore that you want to delete and choose > Delete.

   Warning

   Deleting a metricstore permanently removes all its metrics. This action cannot be undone.

2. In the message that appears, click OK.

Delete metrics

To manually delete metrics, shorten the Data Retention Period for the metricstore.