Pierre Tessier [Sales Engineer|Honeycomb]:
Hi, my name is Pierre Tessier. In this video we’re going to talk about the Honeycomb visualizations, their options, and how to use them.
When we’re discussing visualizations, we need to start with a query. Let’s start with the most basic visualize option of them all, count. So we’re going to go ahead and just put a count on our very basic query and run it. And here we get a count of all the events inside of Honeycomb for this data set. There are many other visualization options as well and they’re all based on the different functions and aggregations that we want to use to render that data. We could perhaps look at the average instead, and in this case, here we need to tell it which measure to use. We’re going to go ahead and go with the duration. When I run this, we’ll now have a line representing the average duration. You can also have more than one visualization per query. It could be against the same measurement using different aggregations or against different measurements themselves. Let’s go ahead and use a Honeycomb heat map against that same duration measurement. When we do this we’ll now have two visuals that are returned for our query.
All your visualization options will respect your where, group by, order by, and limit clauses with your query as well. So in this case here, let’s go ahead and add a group by for app platform. And now when we do this, we’re going to have a line inside of our average duration for every single platform. And below this, we will also have a table representing all of those group options as well. You could hover over each one to highlight inside the visualizations above.
Next, let’s talk about the granularity of the chart. We only have so many pixels on our screen to render the data we have. Honeycomb knows this when it issues a query request. And when the data comes back, we’ll do some form of front end aggregation to make sure we can draw all those raw events onto your screen. In this case here, the time slices are 15 seconds wide. Well we could change that granularity if we want to. I could go with something a little bit more narrow, say 10 seconds, and when I do this we’ll have a little bit more detail in our charts. We could also back it back off and go something much wider, save five minutes. And when you do this, we back off all the detail. This is also akin to doing moving window aggregations as well. We’re going to go put this back to the auto mode, which will select the most optimal view for us. In this case here, the system picked 15 second time slices.
We also have a graph setting here. This allows us to change many different options about the visualizations that we see on the screen right now. We could go ahead and go with stacked graphs, this will render a line chart as a series of stacks instead of individual lines. Let me put this back. Another option here is log scale. And when we go to log scale, this will go ahead and try to bring up the lower values away from the higher values. So we can see that detail, especially at the lower values. And we can see on a heat map here, we have quite a bit of activity that happens at the lower latencies. Going to put this back to linear. You can also change the time axis itself. In my case here, we’re using my browser’s local time zone, but you could change this to be UTC.
Omit missing values is a really fun formula to use, especially if you have sporadic data. I’m going to go ahead and add a filter to this for name equals a popular endpoint that we use. When we do this, you will see the chart lines have a lot more jaggedness going on with them. Going ahead and saying omit missing values will not attempt to drop those lines back down to zero when they’re not reported for that time slice, but connect them to the next reported one. This could be handy especially when you’re getting down to a granularity on your data and you don’t have your data reporting at each one of those time slices and you want to connect those lines together.
On this chart, we have a marker on it representing a build that happened. We can go ahead and remove that marker by selecting the hide markers option. And by doing this it will go away. Let’s go ahead and put it back on. Now you can also go ahead and add markers as you see fit. When you’re hovering a chart, you’ll have a line indicating the time slice itself. At the very top of the line is a button that you could click on to add a marker and we could call this one test marker for example, and we’ll go ahead and create that on to our dataset. This will be available for all visualizations from this point forward.
If you want to create a window, you go ahead and slide and select your window and you’ll be presented with two options. One to add a marker, which you could add it for the window. And the other one to actually zoom in. This is handy if you want to focus in on an area of detail in your chart. When you do the zoom, the system will just render the existing data at whatever time resolution you had. If you’d like to refresh that time resolution, go ahead and hit the run query button and that query will be re-executed at a greater time resolution for the zoom window you selected.
Those are the visualization options inside of Honeycomb. I hope you enjoyed this video. Thank you.
If you see any typos in this text or have any questions, reach out to firstname.lastname@example.org.