You have SLAs you have to meet and SLIs you have to monitor. You have found issues in your system that are being resolved, but you want to verify that the fixes work, or that the issue doesn’t get worse before the fixes arrive. You want to make sure deployments don’t result in regressions. These are all great reasons to monitor your system regularly. But to monitor these very specific things, you need a tool where you can create customized, specific ways to collect data proactively through time, and persist that data in a way that’s easily accessible. That’s where Streams come in.

When you create a Stream that’s based on a query or based on a specific operation, LightStep receives data from your Satellites and stores statistics and example traces related to the Stream to ensure you always have data from 0 to p99.9, including outliers. The Stream view displays statistical time series data and example traces and stores them for as long as your Data Retention policy allows.

Best Practice - Create streams for your ingress operations
Usually, ingress operations for a service are high level enough that they indicate performance problems within a single service, and granular enough that finding the root cause of a performance problem is straightforward.

For example, here’s a Stream for the performance of the iOS service:

The first time series shows latency over a period of time. At the top, you can see the average duration of spans in the p50, p90, p99, and p99.9 buckets. The second series shows the operation rate over time, and the bottom series shows the error percentage rate over time.

When you hover over the latency series, you can view individual spans as dots in a scatterplot. This makes it very easy to see the actual distribution of latency and immediately notice outliers. You can click on any of the dots to open the full trace in Trace view.

Below the time series is the Historical Spans table that displays the spans used to create the stream (it uses the same functionality as the Trace Analysis table in the Explorer view. You can click a span to view the full trace in Trace View . Spans with errors are shown in red.

Once you’re done creating Streams, consider bundling them into dashboards adding alerts to notify team members when thresholds in a Stream are crossed. You can also embed Streams into many third-party applications.

LightStep starts collecting data for Streams once you create them, so when you create a Stream, it will initially be blank. Think about and create Streams before you actually need them!

Create a Stream

You create a Stream based on queries to your span data. LightStep continuously monitors for span data based on that query and saves it as a Stream. You can create a Stream from the Streams view, from your Query in Explorer, or from the Service Directory.

Use our public API to list, retrieve, update, and create Streams. You can also retrieve a specific time series for a Stream.

To create a Stream from the Streams view:

  1. Click Streams from the navigation bar.

  2. Click Create Stream.

  3. Enter a query to base your Stream on and click Run. LightStep runs the query and returns example spans.

  4. If the results are satisfactory, click Create stream. The Stream is added to the list of Streams. By default, the Stream’s name is the query. You can rename it.

  5. To view the Stream, click VIEW STREAM in the banner or click its name in the list.

To create a Stream from Explorer:

  1. Open Explorer from the navigation bar and run any query.

  2. Once the query is run, click Create Stream.

    If the button is grayed out, then a Stream already exists for that query.

  3. To view the Stream, click View Stream.

  4. By default, the Stream takes the name of the query. You can change that to be more descriptive by clicking into the title and entering a name of your choice.

To create a Stream from the Service Directory:

  1. Open the Service Directory from the navigation bar, and choose any service.

  2. To create a Stream for an operation on a service, Click Create Stream for the operation you want to create a Stream for (if the button says View Stream then a stream has already been created for this operation).

  3. To create a Stream for the service, click the Streams tab and click Create Stream.

  4. To view the Stream, click VIEW STREAM in the banner or click View Stream in the list. By default, the Stream’s name is the query. You can rename it.

Regardless of how you create the Stream, you won’t see much data at first because the Satellites have just been told to start collecting it. Over time, the Stream becomes more populated.

Filter Data on a Stream

By default, LightStep displays all data for a Stream from the last 24 hours. You can change that time period. You can also filter the data by errors, percentiles, and duration.

Filtering a Stream also filters the data in the Historical Spans table.

Filter by Time

Use the time period dropdown to choose a different period or enter a custom time period. Click Pick Time Range to enter a custom time period from a calendar. The time series and the Historical Spans table redraw to show spans just from that time period.

Filter by Errors

To view spans with errors, use the Filter dropdown and choose Errors. The series redraws to display only data for spans with errors and the Historical Spans table shows the spans with errors.

The scatterplot of spans does not display when filtering by errors.

Filter by Latency Percentile

By default, LightStep displays lines for the p50, p90, p95, and p99 latency percentiles. You can filter the data to display a specific percentile and above. Use the Filter dropdown and choose Percentile. Enter a value and LightStep redraws the graph and table to show just that percentile and above. You can change the value by clicking it to reopen the dialog.

For example, this data is filtered to show p99.5 and above.

Filter by Duration

By default, LightStep shows all spans, regardless of duration. Use the Filter dropdown, choose Duration, and enter a time period to see where that duration lies in the time series. LightStep redraws the graph to show where that duration lies, and the scatterplot only shows spans at or above that duration. The table shows spans lasting approximately that duration. You can change the value by clicking it to reopen the dialog.

Analyze Span Data

The Historical Spans table shows information from a sample of spans used to create the Stream. By default, the table shows the service; the operation reporting the span, the span’s duration, and the start time. You can add other columns as needed.

When you click on a span, you’re taken to the Trace view where you can see the span in the context of the full trace.

Add All Spans from the Trace to the Table

By default, LightStep only shows spans that match your Stream’s query. But often, you’ll want to see spans that participated in the same traces as the spans returned by your query. For example, if you’re trying to reach a hypothesis, you may want a more broad view of what is going on in a trace, without having to open a bunch of traces yourself.

To see all spans that took part in a trace with your query results, select Show all Spans in Traces.

Add More Columns to the Table

You can add columns that show the span’s tag data to the right of the table by clicking the + icon. As you start typing, LightStep finds tags matching your search.

Filter Data

You can filter the data in the Historical Spans table by Service, Operation, or Tag.

Filtering the table does not affect the data shown in the time series charts.

Group Results

You can group your results to see interesting aggregate data about your spans. You can group by service, operation, or tag.

When you group your results, the table shows each value on the spans for the tag, as well as average latency, error % and span count.

Click on a group to see the spans belonging to that group.

View All Streams

You can view all Streams created for your Organization from the Streams view. Click Streams from the navigation bar.

Streams are listed in alphabetical order. Use the search bar to find Streams quickly.

Rename a Stream

By default, Streams are given the query parameters as the name. You can change that to be more descriptive by clicking the hamburger icon and choosing Rename.

Delete a Stream

You can delete a Stream by clicking the hamburger icon for the Stream on the list view and choosing Delete.

Deleting a Stream also deletes all historical data persisted for that Stream and can’t be undone, so be sure you want to delete it!

Share a Stream

The URL for a Stream is stable and shareable; perfect for alert messages, slack messages, and bot auto-responses. You can share a shortened URL for the Stream by clicking the Share button.

When you integrate LightStep with Slack, you can share a preview of the Stream in any channel of your workspace. Simply paste the URL from the the Share button into a Slack channel. Other Slack members can see the Stream and LightStep users can click View Stream to jump right to that page.

Run a Stream’s Query in Explorer

If you want to view a Stream’s span data in Explorer, you can run the query right from the Stream list view. Click the hamburger icon for the Stream and choose Go to Explorer.

Embed a Stream in Third-Party Apps

You can embed a Stream into any app that supports adding an <iframe> object. By default, the Stream uses the light theme currently displayed in LightStep, but you can embed a dark-themed version instead.

Use our Grafana plugin to view LightStep Streams in Grafana. Once in Grafana, clicking on a LightStep panel takes you back to LightStep where you can view sample traces.

To embed a Stream:

  1. Open the IFrame URL dialog to copy the code for the Stream. You can find the code to embed the Stream in two places:

    From the Stream, at the top, click Embed.

    From a dashboard, click the gear icon for the Stream and choose Get embed URL.

  2. If you want to use a dark version of the Stream, click Dark Theme in the dialog.

  3. Copy the code and paste into the third-party app as needed. For example in Datadog, you can drag an IFrame widget onto a ScreenBoard and paste in the code from LightStep.