Monitoring with Amazon CloudWatch - Amazon Timestream
Using metrics

Monitoring with Amazon CloudWatch

You can monitor Timestream for LiveAnalytics using Amazon CloudWatch, which collects and processes raw data from Timestream for LiveAnalytics into readable, near-real-time metrics. It records these statistics for two weeks so that you can access historical information and gain a better perspective on how your web application or service is performing. By default, Timestream for LiveAnalytics metric data is automatically sent to CloudWatch in 1-minute or 15-minute periods. For more information, see What Is Amazon CloudWatch? in the Amazon CloudWatch User Guide.

Topics

How do I use Timestream for LiveAnalytics metrics?

The metrics reported by Timestream for LiveAnalytics provide information that you can analyze in different ways. The following list shows some common uses for the metrics. These are suggestions to get you started, not a comprehensive list.

How can I?

Relevant metrics
How can I determine if any system errors occurred?

You can monitor SystemErrors to determine whether any requests resulted in a server error code. Typically, this metric should be equal to zero. If it isn't, you might want to investigate.
How can I monitor the amount of data in the memory store?

You can monitor MemoryCumulativeBytesMetered over the specified time period, to monitor the amount of data stored in memory store in bytes. This metric is emitted every hour and you can track the bytes stored at an account as well as at database granularity. The memory store is metered in GB-hour (the cost of storing 1GB of data for one hour). So multiplying the hourly value of MemoryCumulativeBytesMetered with GB-hour pricing in your Region will give you the cost incurred per hour.

Dimensions: Operation (storage), DatabaseName, Metric name
How can I monitor the amount of data in the magnetic store?

You can monitor MagneticCumulativeBytesMetered over the specified time period, to monitor the amount of data stored in magnetic store in bytes. This metric is emitted every hour and you can track the bytes stored at an account as well as at database granularity. The memory store is metered in GB-month (the cost of storing 1GB of data for one month). So multiplying the hourly value of MagneticCumulativeBytesMetered with GB-month pricing in your Region will give you the cost incurred per hour. For example, if the value of MagneticCumulativeBytesMetered is 107374182400 bytes (100GB), then the hourly charge of 1GB of data in magnetic store = (0.03) (us-east-1 pricing) / (30.4*24). Multiplying this value with the MagneticCumulativeBytesMetered in GB will give ~$0.004 for that hour.

Dimensions: Operation (Storage), DatabaseName, Metric name
How can I monitor the data scanned by queries?

You can monitor CumulativeBytesMetered over the specified time period, to monitor the data scanned by queries (in bytes) sent to Timestream for LiveAnalytics. This metric is emitted after the query execution and you can track the data scanned at account and database granularity. You can calculate the query cost for a particular period by multiplying the value of the metric with per GB scanned pricing in your Region. The bytes scanned by scheduled queries are accounted for in this metric.

Dimensions: Operation (Query), DatabaseName, Metric name
How can I monitor the data scanned by scheduled queries?

You can monitor CumulativeBytesMetered over the specified time period, to monitor the data scanned by scheduled queries (in bytes) executed by Timestream for LiveAnalytics. This metric is emitted after the query execution and you can track the data scanned at account and database granularity. You can calculate the query cost for a particular period by multiplying the value of the metric with per GB scanned pricing in your Region.

Note

The bytes metered are also accounted for in the query CumulativeBytesMetered.

Dimensions: Operation (TriggeredScheduledQuery), DatabaseName, Metric name
How can I monitor the number of records ingested?

You can monitor NumberOfRecords over the specified time period to monitor the number of records ingested. You can track the bytes stored at an account as well as at database granularity. You can also use this metric to monitor the writes made by Scheduled Queries when query results are written into a separate table.

When using the WriteRecords API, the metric is emitted for each WriteRecords request, with the CloudWatch Operation dimension being WriteRecords. When using the BatchLoad or ScheduledQuery APIs, the metric is emitted at intervals determined by the service until the task completes. The CloudWatch Operation dimension for this metric is either BatchLoad or ScheduledQuery, depending on which API is used.

Dimensions: Operation (WriteRecords, BatchLoad, or ScheduledQuery), DatabaseName, Metric name
How can I monitor the cost of records ingested?

You can monitor CumulativeBytesMetered to monitor the number of bytes ingested that accrue cost. You can track the bytes stored at an account as well as at database granularity. Ingested records are metered in cumulative bytes. Multiplying the value of CumulativeBytesMetered by Writes pricing in your Region gives you the ingestion cost incurred.

When using the WriteRecords API, this metric is emitted for each WriteRecords request, with the CloudWatch Operation dimension being WriteRecords. When using the BatchLoad or ScheduledQuery API, the metric is emitted at intervals determined by the service until the task completes. The CloudWatch Operation dimension for this metric is BatchLoad or ScheduledQuery depending on which API is used..

Dimensions: Operation (WriteRecords, BatchLoad, or ScheduledQuery), DatabaseName, Metric name
How can I monitor the Timestream Compute Units (TCUs) used in my account?

You can monitor QueryTCU over the specified time period, to monitor the compute units consumed for query workload in the account. This metric is emitted with maximum and minimum compute units for every minute during active query workload from the account.

Units: Count

Valid Statistics: Minimum, Maximum

Metric: ResourceCount

Dimensions: Service: Timestream, Resource: QueryTCU, Type: Resource, Class: OnDemand