Skip to main content
Use Honeycomb to learn and explore your AWS data. Use the Explore Data tab to orient on fields within your logs and metrics data. Explore your data further with Query Builder. For AWS logs and metrics, try some techniques listed below and explore associated resources.

​
How to Find Outliers in Structured Logs

BubbleUp allows you to examine what a subset of logs might have in common. Using structured MySQL slow query logs as an example, BubbleUp can help to discover specific queries that are running slower than expected. In this example, we created a query that visualizes a heatmap on query_time, or VISUALIZE HEATMAP(query_time). BubbleUp allows us to draw a box around logs that have a longer duration than preferred. In the menu that appears, select BubbleUp Outliers.
A screenshot depicting a heatmap on the query_time field with a BubbleUp box drawn around logs with a long duration.
Then, view the BubbleUp charts below the heatmap. Look at which attributes that the logs inside our selected box (in yellow) have in common.
A screenshot depicting the dimensions BubbleUp highlights on the selection.
Now we see that each of those logs was the same DELETE query in the statement field. Using the normalized_query field, we can see the exact shape of the query we need to go optimize.

​
Work with Unstructured Logs

Unstructured logs are received by Honeycomb as events with a verbose message attribute. All unstructured log data is contained within this single attribute, which can make it hard to query. Using Calculated Fields on this attribute can improve the experience and make the logs more useful.

​
Use Calculated Fields on Unstructured Logs

Use calculated fields to parse meaningful information out of the unstructured message attribute. When creating a calculated field, leverage the REG_VALUE operation when parsing. While this experience is better with structured logs, we recommend this as a workaround for unstructured logs.

​
Work With AWS Metrics

To learn more about querying metrics and correlating metrics with other datasets, refer to our Metrics Overview.