We’re excited to announce a complete upgrade to our Metrics front-end. After initially launching the service last December we spent lots of time talking with our users to find out where Metrics could improve. We found that while it was well suited to managing and monitoring scale-out tiers in an IT environment, it was lacking the critical ability to correlate arbitrary data streams on one graph. Rather than just adding a few new screens to the app, we went back to the drawing board and redesigned the interface around this fundamental new capability and took the opportunity to add lots of other improvements. We’ve been hard at work on it ever since, with a select group of users serving as invaluable beta testers (thanks!), and the end result is available to everyone starting today! You can read the details below, or
check out an introductory video.
New Single Metric Pages
Simply pushing data into the service via our
RESTful API or one of the many
language bindings and supported agents is all you need to get started. You can then select any of these metrics and bring up a real time graph for that metric across all sources (e.g. servers, applications, sensors) reporting it, enabling a rapid detection of any abnormally behaving sources.
 |
Configuring alerts/attributes on a metric.
|
On this page you can tweak the configuration of your metric (protip: always set a period and units!) and set threshold-based alarms. We’ve added a much more natural date-picker that allows you to rapidly switch from instantaneous updates to trending over long periods and a gradient of choices in-between. All configuration changes are applied to the visualization immediately so you can tweak things until they are just right prior to saving your changes. We’ve also introduced stream filtering that allows you to specify a pattern that a source name must match (e.g. ‘production’) to remain visible on the graph. Visible series are updated dynamically as you change the filter.
Correlations and Instruments
The most fundamental new feature we’ve added is the ability to create custom correlations and save them as a new resource type we’re introducing called
instruments.
 |
Correlating multiple metrics.
|
The new instrument screen allows you to rapidly select the set of
data streams you want to place on one graph by specifying metric/source pairs. Multiple streams with the same unit type will share a Y-Axis and additional Y-axes are created for different unit types. We also support
wild-card sources that allow you to specify multiple sources at once. Multiple gauge streams specified with a wild-card can be optionally averaged together, aggregated (summed), or broken out individually, and you can dynamically modify this behavior (great for fire-fighting).
New Dashboards
The final piece of our redesign are completely overhauled dashboards. Dashboards are now comprised of instruments and we’ve seamlessly migrated all existing dashboards to the new interface. Creating a new dashboard from scratch takes minutes with our new instrument selection wizard, and you can create as many as you’d like. All instruments auto-refresh and each dashboard has a single date-picker that changes the time-interval in lockstep for all enclosed instruments. Finally, our dashboards intelligently size themselves to whatever display you’ve loaded them on, meaning they look great on anything from an iPad to a wall-mounted 1080p display.
 |
Dashboard on a large plasma display.
|
 |
Dashboard on an iPad.
|
Future Work
This upgrade is just the beginning, we’re committed to ensuring our users have access to a best of breed tool for storing, visualizing, and analyzing all of their time series data. We’ve got a slew of new features lined up and they’ll be rolling out at a much finer granularity over the next few months. If you have any ideas/requests
we’d love to hear them and you should
follow us on Twitter to stay up to date. So what are you waiting for,
start your free trial today!
