How to create Bar chart in CloudWatch dashboard with a chart for each month?

Hi, you are right in your understanding that to plot a bar chart with each month being a distinct bar, you need to have month as a dimension of the chart. If what you want to plot is a custom metric, you can indeed add a dimension to it. If however the input is a standard metric, you cannot add a dimension to it and you need to duplicate the metric to a custom metric for that purpose, which implies paying for a custom metric (unless you haven’t exhausted the free tier yet). An alternate solution would be to use multi datasource querying with a custom lambda who runs a GetMetricData behinds the scene and reorganizes the data by month on the fly, which is not too different from what you would do anyway if you need to duplicate a vended metric to a custom metric. The benefit is however is there would be no upfront cost for custom metric(s), only GetMetricData API and Lambda execution costs when you plot/refresh the widget. The other alternative would be an entirely custom widget, but you would have to source the data from the metrics anyway, so it would not bring any cost benefit. It would probably add a bit of complexity to your lambda, although the benefit would be complete flexibility in your layout so you could supercharge your bar chart with other data or interactions than a standard graph.

Creating a bar chart in CloudWatch with a bar for each month over the last 6 months is indeed challenging due to CloudWatch's limitations in handling time-based aggregations directly. However, there are a few approaches you can consider:

1. Using CloudWatch Metrics Insights: This feature allows you to query and analyze your metrics data using SQL-like queries. You might be able to create a query that aggregates your TranscodingOutputDuration metric by month. However, this would require writing a custom query and may not directly produce a bar chart visualization.

2. Custom Lambda Function: You could create a Lambda function that retrieves the metric data for the last 6 months, aggregates it by month, and then publishes these aggregated values as new custom metrics. These custom metrics could then be used to create the bar chart in your dashboard.

3. Using Amazon QuickSight: If you need more advanced visualizations, you might consider using Amazon QuickSight. It can connect to CloudWatch metrics and provide more flexible options for creating time-based visualizations.

4. Amazon Athena with CloudWatch Logs Insights: If your metric data is available in CloudWatch Logs, you could use CloudWatch Logs Insights to query the data and then visualize it using Amazon Athena and QuickSight.

5. Custom Widget: You could create a custom widget for your CloudWatch dashboard. This would involve writing a Lambda function that retrieves and processes the metric data, and then renders it as a bar chart using a charting library.

Regarding your approach of adding a "month" dimension to the metric, while this can work for manual creation, it's not ideal for automated updates as you've discovered. The need to redeploy monthly makes it impractical for long-term use.

Given your requirements, CloudWatch alone may not be the best fit for this specific visualization need. You might need to combine it with other AWS services or consider using a more specialized data visualization tool that can pull data from CloudWatch metrics.

If you decide to stick with CloudWatch, the custom widget approach might be your best option as it allows for more flexibility in data processing and visualization while still keeping everything within the CloudWatch dashboard ecosystem.
Sources
Creating a custom widget for a CloudWatch dashboard - Amazon CloudWatch
Create a CloudWatch graph with a search expression - Amazon CloudWatch