Amazon CloudWatch Construct Library

Metric objects

Metric objects represent a metric that is emitted by AWS services or your own application, such as CPUUsage, FailureCount or Bandwidth.

Metric objects can be constructed directly or are exposed by resources as attributes. Resources that expose metrics will have functions that look like metricXxx() which will return a Metric object, initialized with defaults that make sense.

For example, lambda.Function objects have the fn.metricErrors() method, which represents the amount of errors reported by that Lambda function:

# fn: lambda.Function

errors = fn.metric_errors()

Metric objects can be account and region aware. You can specify account and region as properties of the metric, or use the metric.attachTo(Construct) method. metric.attachTo() will automatically copy the region and account fields of the Construct, which can come from anywhere in the Construct tree.

You can also instantiate Metric objects to reference any published metric that’s not exposed using a convenience method on the CDK construct. For example:

hosted_zone = route53.HostedZone(self, "MyHostedZone", zone_name="example.org")
metric = cloudwatch.Metric(
    namespace="AWS/Route53",
    metric_name="DNSQueries",
    dimensions_map={
        "HostedZoneId": hosted_zone.hosted_zone_id
    }
)

Instantiating a new Metric object

If you want to reference a metric that is not yet exposed by an existing construct, you can instantiate a Metric object to represent it. For example:

metric = cloudwatch.Metric(
    namespace="MyNamespace",
    metric_name="MyMetric",
    dimensions_map={
        "ProcessingStep": "Download"
    }
)

Metric Math

Math expressions are supported by instantiating the MathExpression class. For example, a math expression that sums two other metrics looks like this:

# fn: lambda.Function


all_problems = cloudwatch.MathExpression(
    expression="errors + throttles",
    using_metrics={
        "errors": fn.metric_errors(),
        "throttles": fn.metric_throttles()
    }
)

You can use MathExpression objects like any other metric, including using them in other math expressions:

# fn: lambda.Function
# all_problems: cloudwatch.MathExpression


problem_percentage = cloudwatch.MathExpression(
    expression="(problems / invocations) * 100",
    using_metrics={
        "problems": all_problems,
        "invocations": fn.metric_invocations()
    }
)

Search Expressions

Math expressions also support search expressions. For example, the following search expression returns all CPUUtilization metrics that it finds, with the graph showing the Average statistic with an aggregation period of 5 minutes:

cpu_utilization = cloudwatch.MathExpression(
    expression="SEARCH('{AWS/EC2,InstanceId} MetricName=\"CPUUtilization\"', 'Average', 300)",

    # Specifying '' as the label suppresses the default behavior
    # of using the expression as metric label. This is especially appropriate
    # when using expressions that return multiple time series (like SEARCH()
    # or METRICS()), to show the labels of the retrieved metrics only.
    label=""
)

Cross-account and cross-region search expressions are also supported. Use the searchAccount and searchRegion properties to specify the account and/or region to evaluate the search expression against.

Aggregation

To graph or alarm on metrics you must aggregate them first, using a function like Average or a percentile function like P99. By default, most Metric objects returned by CDK libraries will be configured as Average over 300 seconds (5 minutes). The exception is if the metric represents a count of discrete events, such as failures. In that case, the Metric object will be configured as Sum over 300 seconds, i.e. it represents the number of times that event occurred over the time period.

If you want to change the default aggregation of the Metric object (for example, the function or the period), you can do so by passing additional parameters to the metric function call:

# fn: lambda.Function


minute_error_rate = fn.metric_errors(
    statistic=cloudwatch.Stats.AVERAGE,
    period=Duration.minutes(1),
    label="Lambda failure rate"
)

The statistic field accepts a string; the cloudwatch.Stats object has a number of predefined factory functions that help you constructs strings that are appropriate for CloudWatch. The metricErrors function also allows changing the metric label or color, which will be useful when embedding them in graphs (see below).

Rates versus Sums

The reason for using Sum to count discrete events is that some events are emitted as either 0 or 1 (for example Errors for a Lambda) and some are only emitted as 1 (for example NumberOfMessagesPublished for an SNS topic).

In case 0-metrics are emitted, it makes sense to take the Average of this metric: the result will be the fraction of errors over all executions.

If 0-metrics are not emitted, the Average will always be equal to 1, and not be very useful.

In order to simplify the mental model of Metric objects, we default to aggregating using Sum, which will be the same for both metrics types. If you happen to know the Metric you want to alarm on makes sense as a rate (Average) you can always choose to change the statistic.

Available Aggregation Statistics

For your metrics aggregation, you can use the following statistics:

Statistic	Short format	Long format	Factory name
SampleCount (n)	❌	❌	`Stats.SAMPLE_COUNT`
Average (avg)	❌	❌	`Stats.AVERAGE`
Sum	❌	❌	`Stats.SUM`
Minimum (min)	❌	❌	`Stats.MINIMUM`
Maximum (max)	❌	❌	`Stats.MAXIMUM`
Interquartile mean (IQM)	❌	❌	`Stats.IQM`
Percentile (p)	`p99`	❌	`Stats.p(99)`
Winsorized mean (WM)	`wm99` = `WM(:99%)`	`WM(x:y) \\| WM(x%:y%) \\| WM(x%:) \\| WM(:y%)`	`Stats.wm(10, 90)`
Trimmed count (TC)	`tc99` = `TC(:99%)`	`TC(x:y) \\| TC(x%:y%) \\| TC(x%:) \\| TC(:y%)`	`Stats.tc(10, 90)`
Trimmed sum (TS)	`ts99` = `TS(:99%)`	`TS(x:y) \\| TS(x%:y%) \\| TS(x%:) \\| TS(:y%)`	`Stats.ts(10, 90)`
Percentile rank (PR)	❌	`PR(x:y) \\| PR(x:) \\| PR(:y)`	`Stats.pr(10, 5000)`

The most common values are provided in the cloudwatch.Stats class. You can provide any string if your statistic is not in the class.

# hosted_zone: route53.HostedZone


cloudwatch.Metric(
    namespace="AWS/Route53",
    metric_name="DNSQueries",
    dimensions_map={
        "HostedZoneId": hosted_zone.hosted_zone_id
    },
    statistic=cloudwatch.Stats.SAMPLE_COUNT,
    period=Duration.minutes(5)
)

cloudwatch.Metric(
    namespace="AWS/Route53",
    metric_name="DNSQueries",
    dimensions_map={
        "HostedZoneId": hosted_zone.hosted_zone_id
    },
    statistic=cloudwatch.Stats.p(99),
    period=Duration.minutes(5)
)

cloudwatch.Metric(
    namespace="AWS/Route53",
    metric_name="DNSQueries",
    dimensions_map={
        "HostedZoneId": hosted_zone.hosted_zone_id
    },
    statistic="TS(7.5%:90%)",
    period=Duration.minutes(5)
)

Labels

Metric labels are displayed in the legend of graphs that include the metrics.

You can use dynamic labels to show summary information about the displayed time series in the legend. For example, if you use:

# fn: lambda.Function


minute_error_rate = fn.metric_errors(
    statistic=cloudwatch.Stats.SUM,
    period=Duration.hours(1),

    # Show the maximum hourly error count in the legend
    label="[max: ${MAX}] Lambda failure rate"
)

As the metric label, the maximum value in the visible range will be shown next to the time series name in the graph’s legend.

If the metric is a math expression producing more than one time series, the maximum will be individually calculated and shown for each time series produce by the math expression.

Alarms

Alarms can be created on metrics in one of two ways. Either create an Alarm object, passing the Metric object to set the alarm on:

# fn: lambda.Function


cloudwatch.Alarm(self, "Alarm",
    metric=fn.metric_errors(),
    threshold=100,
    evaluation_periods=2
)

Alternatively, you can call metric.createAlarm():

# fn: lambda.Function


fn.metric_errors().create_alarm(self, "Alarm",
    threshold=100,
    evaluation_periods=2
)

The most important properties to set while creating an Alarms are:

threshold: the value to compare the metric against.
comparisonOperator: the comparison operation to use, defaults to metric >= threshold.
evaluationPeriods: how many consecutive periods the metric has to be breaching the threshold for the alarm to trigger.

To create a cross-account alarm, make sure you have enabled cross-account functionality in CloudWatch. Then, set the account property in the Metric object either manually or via the metric.attachTo() method.

Please note that it is not possible to:

Create a cross-Account alarm that has evaluateLowSampleCountPercentile: "ignore". The reason is that the only way to pass an AccountID is to use the Metrics field of the Alarm resource. If we use the Metrics field, the CloudWatch event that is used to evaluate the Alarm doesn’t have a SampleCount field anymore (”When CloudWatch evaluates alarms, periods are aggregated into single data points”). The result is that the Alarm cannot evaluate at all.
Create a cross-Region alarm (source).

Alarm Actions

To add actions to an alarm, use the integration classes from the aws-cdk-lib/aws-cloudwatch-actions package. For example, to post a message to an SNS topic when an alarm breaches, do the following:

import aws_cdk.aws_cloudwatch_actions as cw_actions
# alarm: cloudwatch.Alarm


topic = sns.Topic(self, "Topic")
alarm.add_alarm_action(cw_actions.SnsAction(topic))

Notification formats

Alarms can be created in one of two “formats”:

With “top-level parameters” (these are the classic style of CloudWatch Alarms).
With a list of metrics specifications (these are the modern style of CloudWatch Alarms).

For backwards compatibility, CDK will try to create classic, top-level CloudWatch alarms as much as possible, unless you are using features that cannot be expressed in that format. Features that require the new-style alarm format are:

Metric math
Cross-account metrics
Labels

The difference between these two does not impact the functionality of the alarm in any way, except that the format of the notifications the Alarm generates is different between them. This affects both the notifications sent out over SNS, as well as the EventBridge events generated by this Alarm. If you are writing code to consume these notifications, be sure to handle both formats.

Composite Alarms

Composite Alarms can be created from existing Alarm resources.

# alarm1: cloudwatch.Alarm
# alarm2: cloudwatch.Alarm
# alarm3: cloudwatch.Alarm
# alarm4: cloudwatch.Alarm


alarm_rule = cloudwatch.AlarmRule.any_of(
    cloudwatch.AlarmRule.all_of(
        cloudwatch.AlarmRule.any_of(alarm1,
            cloudwatch.AlarmRule.from_alarm(alarm2, cloudwatch.AlarmState.OK), alarm3),
        cloudwatch.AlarmRule.not(cloudwatch.AlarmRule.from_alarm(alarm4, cloudwatch.AlarmState.INSUFFICIENT_DATA))),
    cloudwatch.AlarmRule.from_boolean(False))

cloudwatch.CompositeAlarm(self, "MyAwesomeCompositeAlarm",
    alarm_rule=alarm_rule
)

Actions Suppressor

If you want to disable actions of a Composite Alarm based on a certain condition, you can use Actions Suppression.

# alarm1: cloudwatch.Alarm
# alarm2: cloudwatch.Alarm
# on_alarm_action: cloudwatch.IAlarmAction
# on_ok_action: cloudwatch.IAlarmAction
# actions_suppressor: cloudwatch.Alarm


alarm_rule = cloudwatch.AlarmRule.any_of(alarm1, alarm2)

my_composite_alarm = cloudwatch.CompositeAlarm(self, "MyAwesomeCompositeAlarm",
    alarm_rule=alarm_rule,
    actions_suppressor=actions_suppressor
)
my_composite_alarm.add_alarm_action(on_alarm_action)
my_composite_alarm.add_ok_action(on_ok_action)

In the provided example, if actionsSuppressor is in ALARM state, onAlarmAction won’t be triggered even if myCompositeAlarm goes into ALARM state. Similar, if actionsSuppressor is in ALARM state and myCompositeAlarm goes from ALARM into OK state, onOkAction won’t be triggered.

A note on units

In CloudWatch, Metrics datums are emitted with units, such as seconds or bytes. When Metric objects are given a unit attribute, it will be used to filter the stream of metric datums for datums emitted using the same unit attribute.

In particular, the unit field is not used to rescale datums or alarm threshold values (for example, it cannot be used to specify an alarm threshold in Megabytes if the metric stream is being emitted as bytes).

You almost certainly don’t want to specify the unit property when creating Metric objects (which will retrieve all datums regardless of their unit), unless you have very specific requirements. Note that in any case, CloudWatch only supports filtering by unit for Alarms, not in Dashboard graphs.

Please see the following GitHub issue for a discussion on real unit calculations in CDK: https://github.com/aws/aws-cdk/issues/5595

Dashboards

Dashboards are set of Widgets stored server-side which can be accessed quickly from the AWS console. Available widgets are graphs of a metric over time, the current value of a metric, or a static piece of Markdown which explains what the graphs mean.

The following widgets are available:

GraphWidget – shows any number of metrics on both the left and right vertical axes.
AlarmWidget – shows the graph and alarm line for a single alarm.
SingleValueWidget – shows the current value of a set of metrics.
TextWidget – shows some static Markdown.
AlarmStatusWidget – shows the status of your alarms in a grid view.

Graph widget

A graph widget can display any number of metrics on either the left or right vertical axis:

# dashboard: cloudwatch.Dashboard
# execution_count_metric: cloudwatch.Metric
# error_count_metric: cloudwatch.Metric


dashboard.add_widgets(cloudwatch.GraphWidget(
    title="Executions vs error rate",

    left=[execution_count_metric],

    right=[error_count_metric.with(
        statistic=cloudwatch.Stats.AVERAGE,
        label="Error rate",
        color=cloudwatch.Color.GREEN
    )]
))

Using the methods addLeftMetric() and addRightMetric() you can add metrics to a graph widget later on.

Graph widgets can also display annotations attached to the left or right y-axis or the x-axis.

# dashboard: cloudwatch.Dashboard


dashboard.add_widgets(cloudwatch.GraphWidget(
    # ...

    left_annotations=[cloudwatch.HorizontalAnnotation(value=1800, label=Duration.minutes(30).to_human_string(), color=cloudwatch.Color.RED), cloudwatch.HorizontalAnnotation(value=3600, label="1 hour", color="#2ca02c")
    ],
    vertical_annotations=[cloudwatch.VerticalAnnotation(date="2022-10-19T00:00:00Z", label="Deployment", color=cloudwatch.Color.RED)
    ]
))

The graph legend can be adjusted from the default position at bottom of the widget.

# dashboard: cloudwatch.Dashboard


dashboard.add_widgets(cloudwatch.GraphWidget(
    # ...

    legend_position=cloudwatch.LegendPosition.RIGHT
))

The graph can publish live data within the last minute that has not been fully aggregated.

# dashboard: cloudwatch.Dashboard


dashboard.add_widgets(cloudwatch.GraphWidget(
    # ...

    live_data=True
))

The graph view can be changed from default ‘timeSeries’ to ‘bar’ or ‘pie’.

# dashboard: cloudwatch.Dashboard


dashboard.add_widgets(cloudwatch.GraphWidget(
    # ...

    view=cloudwatch.GraphWidgetView.BAR
))

The start and end properties can be used to specify the time range for each graph widget independently from those of the dashboard. The parameters can be specified at GraphWidget, GaugeWidget, and SingleValueWidget.

# dashboard: cloudwatch.Dashboard


dashboard.add_widgets(cloudwatch.GraphWidget(
    # ...

    start="-P7D",
    end="2018-12-17T06:00:00.000Z"
))

Table Widget

A TableWidget can display any number of metrics in tabular form.

# dashboard: cloudwatch.Dashboard
# execution_count_metric: cloudwatch.Metric


dashboard.add_widgets(cloudwatch.TableWidget(
    title="Executions",
    metrics=[execution_count_metric]
))

The layout property can be used to invert the rows and columns of the table. The default cloudwatch.TableLayout.HORIZONTAL means that metrics are shown in rows and datapoints in columns. cloudwatch.TableLayout.VERTICAL means that metrics are shown in columns and datapoints in rows.

# dashboard: cloudwatch.Dashboard


dashboard.add_widgets(cloudwatch.TableWidget(
    # ...

    layout=cloudwatch.TableLayout.VERTICAL
))

The summary property allows customizing the table to show summary columns (columns sub property), whether to make the summary columns sticky remaining in view while scrolling (sticky sub property), and to optionally only present summary columns (hideNonSummaryColumns sub property).

# dashboard: cloudwatch.Dashboard


dashboard.add_widgets(cloudwatch.TableWidget(
    # ...

    summary=cloudwatch.TableSummaryProps(
        columns=[cloudwatch.TableSummaryColumn.AVERAGE],
        hide_non_summary_columns=True,
        sticky=True
    )
))

The thresholds property can be used to highlight cells with a color when the datapoint value falls within the threshold.

# dashboard: cloudwatch.Dashboard


dashboard.add_widgets(cloudwatch.TableWidget(
    # ...

    thresholds=[
        cloudwatch.TableThreshold.above(1000, cloudwatch.Color.RED),
        cloudwatch.TableThreshold.between(500, 1000, cloudwatch.Color.ORANGE),
        cloudwatch.TableThreshold.below(500, cloudwatch.Color.GREEN)
    ]
))

The showUnitsInLabel property can be used to display what unit is associated with a metric in the label column.

# dashboard: cloudwatch.Dashboard


dashboard.add_widgets(cloudwatch.TableWidget(
    # ...

    show_units_in_label=True
))

The fullPrecision property can be used to show as many digits as can fit in a cell, before rounding.

# dashboard: cloudwatch.Dashboard


dashboard.add_widgets(cloudwatch.TableWidget(
    # ...

    full_precision=True
))

Single value widget

A single-value widget shows the latest value of a set of metrics (as opposed to a graph of the value over time):

# dashboard: cloudwatch.Dashboard
# visitor_count: cloudwatch.Metric
# purchase_count: cloudwatch.Metric


dashboard.add_widgets(cloudwatch.SingleValueWidget(
    metrics=[visitor_count, purchase_count]
))

Show as many digits as can fit, before rounding.

# dashboard: cloudwatch.Dashboard


dashboard.add_widgets(cloudwatch.SingleValueWidget(
    metrics=[],

    full_precision=True
))

Sparkline allows you to glance the trend of a metric by displaying a simplified linegraph below the value. You can’t use sparkline: true together with setPeriodToTimeRange: true

# dashboard: cloudwatch.Dashboard


dashboard.add_widgets(cloudwatch.SingleValueWidget(
    metrics=[],

    sparkline=True
))

Period allows you to set the default period for the widget:

# dashboard: cloudwatch.Dashboard


dashboard.add_widgets(cloudwatch.SingleValueWidget(
    metrics=[],

    period=Duration.minutes(15)
))

Alarm Status widget

An alarm status widget displays instantly the status of any type of alarms and gives the ability to aggregate one or more alarms together in a small surface.

# dashboard: cloudwatch.Dashboard
# error_alarm: cloudwatch.Alarm


dashboard.add_widgets(
    cloudwatch.AlarmStatusWidget(
        alarms=[error_alarm]
    ))

An alarm status widget only showing firing alarms, sorted by state and timestamp:

# dashboard: cloudwatch.Dashboard
# error_alarm: cloudwatch.Alarm


dashboard.add_widgets(cloudwatch.AlarmStatusWidget(
    title="Errors",
    alarms=[error_alarm],
    sort_by=cloudwatch.AlarmStatusWidgetSortBy.STATE_UPDATED_TIMESTAMP,
    states=[cloudwatch.AlarmState.ALARM]
))

Dashboard Layout

The widgets on a dashboard are visually laid out in a grid that is 24 columns wide. Normally you specify X and Y coordinates for the widgets on a Dashboard, but because this is inconvenient to do manually, the library contains a simple layout system to help you lay out your dashboards the way you want them to.

Widgets have a width and height property, and they will be automatically laid out either horizontally or vertically stacked to fill out the available space.

Widgets are added to a Dashboard by calling add(widget1, widget2, ...). Widgets given in the same call will be laid out horizontally. Widgets given in different calls will be laid out vertically. To make more complex layouts, you can use the following widgets to pack widgets together in different ways:

Column: stack two or more widgets vertically.
Row: lay out two or more widgets horizontally.
Spacer: take up empty space

Interval duration for dashboard

Interval duration for metrics in dashboard. You can specify defaultInterval with the relative time(eg. 7 days) as Duration.days(7).

import aws_cdk.aws_cloudwatch as cw


dashboard = cw.Dashboard(self, "Dash",
    default_interval=Duration.days(7)
)

Here, the dashboard would show the metrics for the last 7 days.

Dashboard variables

Dashboard variables are a convenient way to create flexible dashboards that display different content depending on the value of an input field within a dashboard. They create a dashboard on which it’s possible to quickly switch between different Lambda functions, Amazon EC2 instances, etc.

You can learn more about Dashboard variables in the Amazon Cloudwatch User Guide

There are two types of dashboard variables available: a property variable and a pattern variable.

Property variables can change any JSON property in the JSON source of a dashboard like region. It can also change the dimension name for a metric.
Pattern variables use a regular expression pattern to change all or part of a JSON property.

A use case of a property variable is a dashboard with the ability to toggle the region property to see the same dashboard in different regions:

import aws_cdk.aws_cloudwatch as cw


dashboard = cw.Dashboard(self, "Dash",
    default_interval=Duration.days(7),
    variables=[cw.DashboardVariable(
        id="region",
        type=cw.VariableType.PROPERTY,
        label="Region",
        input_type=cw.VariableInputType.RADIO,
        value="region",
        values=cw.Values.from_values(cw.VariableValue(label="IAD", value="us-east-1"), label="DUB", value="us-west-2"),
        default_value=cw.DefaultValue.value("us-east-1"),
        visible=True
    )]
)

This example shows how to change region everywhere, assuming the current dashboard is showing region us-east-1 already, by using pattern variable

import aws_cdk.aws_cloudwatch as cw


dashboard = cw.Dashboard(self, "Dash",
    default_interval=Duration.days(7),
    variables=[cw.DashboardVariable(
        id="region2",
        type=cw.VariableType.PATTERN,
        label="RegionPattern",
        input_type=cw.VariableInputType.INPUT,
        value="us-east-1",
        default_value=cw.DefaultValue.value("us-east-1"),
        visible=True
    )]
)

The following example generates a Lambda function variable, with a radio button for each function. Functions are discovered by a metric query search. The values with cw.Values.fromSearchComponents indicates that the values will be populated from FunctionName values retrieved from the search expression {AWS/Lambda,FunctionName} MetricName=\"Duration\". The defaultValue with cw.DefaultValue.FIRST indicates that the default value will be the first value returned from the search.

import aws_cdk.aws_cloudwatch as cw


dashboard = cw.Dashboard(self, "Dash",
    default_interval=Duration.days(7),
    variables=[cw.DashboardVariable(
        id="functionName",
        type=cw.VariableType.PATTERN,
        label="Function",
        input_type=cw.VariableInputType.RADIO,
        value="originalFuncNameInDashboard",
        # equivalent to cw.Values.fromSearch('{AWS/Lambda,FunctionName} MetricName=\"Duration\"', 'FunctionName')
        values=cw.Values.from_search_components(
            namespace="AWS/Lambda",
            dimensions=["FunctionName"],
            metric_name="Duration",
            populate_from="FunctionName"
        ),
        default_value=cw.DefaultValue.FIRST,
        visible=True
    )]
)

You can add a variable after object instantiation with the method dashboard.addVariable().