Share Context With Collaboration: Suggested Query Boards

Nobody knows your services/infra better than you, not even Honeycomb.

If there’s one maxim for Honeycomb, it’s that context is king. Context determines the questions you can ask. The only way to make complex systems truly tractable is to make all questions possible. Ergo: more context. Context everywhere. Context coming out of the walls. You should be awash in context.

We talk a lot about context when it comes to the events you send, about how they should be as wide and rich as possible. But there’s another place where context can make or break your team’s experience using Honeycomb: in the UI itself. Metacontext, if you will.

This is the first in a series of posts detailing ways to both improve your team’s experience with Honeycomb, but to also leverage Honeycomb’s collaboration features to distribute context from individual members of the team across your entire team.

cartoon of a dog showing a fish to another dog and saying "i don't know what this is but it seems useful!"

New in Honeycomb: Suggested Queries

For several common technologies, Honeycomb now provides Suggested Queries when you land on the query page:

suggested queries shown for a dataset

We have a pretty good collection of recognized dataset schemas (and we keep adding to it), but chances are also pretty good that you’ve never even seen that feature. Custom instrumentation just doesn’t get recognized.

The purpose of Suggested Queries isn’t to enumerate specific problems. That’s one step away from dashboards, and not the way we believe people should be using Honeycomb. Here’s how we recommend using them:

  1. Identifying common jumping off points to start exploring when you don’t even know what the problem is.
  2. Onboarding members on your team who may have never even looked at the dataset before (or to remind yourself, after you’ve been away for too long :)).

Regardless of the amount of time we spend recognizing dataset schema here at Honeycomb, we will never have the context that your team has built up in the process of instrumenting and running your services to give you really useful suggestions. Maybe in some grand AI future, but not now. Right now, the best suggestions come from those with the most context – and that’s you.

We recently flipped the switch enabling Suggested Query Boards. This feature allows you to specify a Board as the source of suggestions for a dataset.

For example, check out this Board:

screenshot of suggested queries for a dataset

This Board contains a few queries for our “poodle” dataset. If we go to that dataset’s overview page, you can see the new control for selecting a Suggested Query Board. Selecting that Board is easy:

gif of selecting a Board

And here’s what it looks like on the blank query page. The title is a link directly to the Board:

screenshot of the suggested queries

There are some restrictions on which Boards can be used, and which queries within them will show up, but the biggest ones are:

  1. The Board must be public.
  2. Queries must be named. Unnamed queries don’t really work well as suggestions.
  3. Though not a requirement, add descriptions! The query name is the what, and the description is your chance to explain (succinctly!) the why. Don’t skimp. Your teammates will thank you.
  4. Only the first 4 queries matching the dataset are shown.

We’ve been hustling to add Suggested Queries Boards to our many internal datasets, some of which only have a couple of folks who query them regularly. Those in particular benefit from Suggested Queries, since they’re more likely to fall out of the brains of people who do query them, and aren’t queried often by the rest of the team.

Multiple datasets can pull suggested queries from the same Board, so you could have a single “Suggested Queries Board”, for multiple datasets, or one board per dataset. Whatever works for you.

Give it a whirl and show us how you use suggested queries to level up your team–we might send you an “I Test in Prod” t-shirt 🙂

If you’re ready to get started with Honeycomb, sign up for a demo today!