Adam Kinniburgh
VP Innovation, SquaredUp
A dive into health roll-up
SquaredUp's awesome dashboards sit on top of some very powerful technology. In this post, I'll explain just what that means and how it can benefit you.
If you want me to show you instead, just click play below.
What is a health roll-up?
Put simply, health roll-up means taking the health from one "object", and using it to influence the health of another, typically something sitting higher up in your hierarchy. Health, that rolls up.
In SquaredUp's context, we're talking about three key object types. Monitors, Dashboards, and Workspaces.
When you create a tile on a dashboard, you have the option to enable Monitoring. This allows you to create a health state for that piece of data, which we represent as a Monitor. Maybe it's a threshold on a metric i.e. when your response time goes over 100ms. Perhaps its a state monitor, used when another tool is already telling us something.
Roll-up starts right from there. Your tile's health influences the health of the dashboard it lives on, and that dashboard then influences the health of the workspace it lives in. At the time of writing, our health model uses the state of the worst member in that roll-up. So if you have a dashboard with 2 tiles and one goes red, that "worst state" is what the dashboard inherits. This continues up the chain.
The first win
WIth a few dashboards, in a few workspaces, with a few monitors enabled, the first win is that your global home view will light up like a Christmas tree. You can set the view to show Monitors, Dashboards, or Workspaces, and can then further filter this view by Tags, Health, or Types.
So with just a few monitors turned on, you're already a step towards that coveted "Executive View", uncluttered by busy line graphs and metrics, just the simple red, amber, green boxes they crave in the boardroom.
The filter preferences on this view are per user, so while your view might be all of the micro-service monitors from workspaces tagged up with your name, your boss might want to see all the application workspaces, and the exec team might want just department health.
Building a working map
Monitor to Dashboard to Workspace is just the start though. That shows health in a very focussed way, but there's no structure yet. What if you want the health of a micro-service to impact the health of an upstream caller, or the performance of a sales rep to influence the health of their team... Now we're in the world of Dependencies.
SquaredUp was built to make dependency mapping simple and flexible. There are a few different ways to use all these juicy health states depending on what you want to achieve.
I just want to see the health of another team's service
Easy... when building your dashboard, you simply choose the SquaredUp Health data stream, then scope to the monitors, dashboards, or workspaces of your choice. You'll get a simple block tile showing the state of those objects.
I want a simple roll-up as well
You can get a simple roll-up in this case by enabling a monitor on your new tiles, choosing the State option. Now, if one of the objects you've chosen becomes unhealthy, your tile will become unhealthy too, as will the dashboard and the workspace. That downstream health from somewhere in the ether is now rolling up into your workspace. Neat!
As a quick note at this point, now that you've got monitoring in your workspace, you can enable Notifications and SquaredUp can tell you via Slack, Teams, Zapier, webhooks, and more, when something isn't right. No need to wait for the other team to let you know they're in trouble.
I want the whole fancy map
Well I don't blame you...
The map is powered by Workspace Dependencies. These are slightly different to just using health states on your dashboards. Modelling your whole application, team, or even company is just as easy though.
From your workspace, go to the Map panel. You'll see your lonely old workspace all by itself. If you hover over it, you'll see a + icon appear below. SquaredUp maps from the top down (at the time of writing), so it's easiest to start at the top.
For example, maybe you have an Engineering workspace. Starting there, you might map some downstream dependencies to the workspaces for each of your squads. From their workspaces, they might map down to the apps they're responsible for. From the apps, it's the micro-services, and maybe further beyond.
By the time everyone has created the links to the things they depend on, you'll get a killer view from your organization's home. Simply toggle from Tile to Map, and voila. From chaos comes order and beauty!
With all of these dependencies in place, the health of each workspace will be inherited from all of those below. Issues with your lowest level services can be seen more easily, and why they're a problem can finally be understood.
I want all of the above, in my inbox
Easy win! Any time you enable monitoring anywhere in SquaredUp, you're creating something that can trigger a Notification. We currently support sending notifications to email, Slack, Teams, ServiceNow, Zapier, and Webhooks.
Just jump into the Monitors tab from your chosen Workspace. You can create the destinations you need, and have fully granular control over what gets sent to who!
In addition to triggering based on Monitor conditions, our Notification API can be used to trigger a message whenever you want. Here at SquaredUp, we have notifications that include a whole dashboard being delivered to Slack channels at 9am each day. A great way to speed up those team stand-ups.
It's not all on your shoulders
The real beauty with our health model is that for most customers, it "just happens". With the top-down mapping approach, it's in everyone's interest to identify all of the things that impact them. By making it simple for everyone to just point-and-click those relationships into existence, the entire organization benefits from that collective knowledge, all seamlessly correlated and colour-coded. No one person needs to achieve this on their own.
Health roll-up, made beautifully simple.
Adam Kinniburgh
VP Innovation, SquaredUp
