To blog |

Product Updates – 18 Feb 2019

February 18, 2019 by Ram Iyengar Filed under: Plumbr Product Updates

In an effort to improve the benefits that Plumbr can offer, we are introducing important changes to several key areas of the product. Please pay attention to all the information that follows. It may affect your current workflows. This is the list of the changes we are making:

  • Shifting key performance metric to time
  • Percentiles to replace classification of interactions or API calls based on duration
  • Retiring the use of time-based thresholds
  • Changes to dashboard and summary widgets to support the new approach
  • Availability and performance are now exposed separately
  • Interactions and API calls in “Stuck” status are now shifted to performance
  • Individual Interactions provided in the context of a specific error/bottleneck

Now for some details about each change:

The right perspective on performance!

The most important change that we are announcing today is a fundamental shift in how the performance of the application is exposed. We will now be using response times as the metric  for communicating the performance characteristics of the application. Previously, we were exposing this information as a count of affected users or API calls. Since the industry-wide standard for communicating performance metrics is based on response times, we have decided to make a change in Plumbr.

Update to the dashboard. Note that it shows duration for performance.
All your applications will now show information based on time.

More focus on percentiles

Percentiles are the appropriate way to quantify the performance of web applications. So far, we exposed them in a limited manner, but with this update, percentiles will be used in a widespread manner within the product.

As a result, lists of applications, interactions, services, API calls, etc, will reflect the median (50th), 90th, and 99th percentile.

Performance summary of an application


List of applications exposing percentiles

An Updated performance summary for your applications

As a result of this update, the performance summary of your applications become much more actionable and insightful. By introducing timing in the context of sessions, session duration, interactions, and waiting time, performance engineering for your applications takes on much more meaning. You can make informed decisions about where to invest time, money and effort.

Performance summary of an application
Using percentiles to characterize performance

Retiring the use of Thresholds

This transition renders the concept of thresholds obsolete. We will be removing the ability to apply a threshold for your applications/services/interactions to be flagged as slow. If you are currently using present thresholds or custom thresholds, please get in touch with our Customer Success team to understand the right path to migrate away.

More meaningful dashboards

When monitoring browser interactions, the Plumbr dashboard will now show the following tiles:

  • Total user interactions monitored
  • Number of users facing failures
  • Cumulative waiting time experienced by users
Dashboard showing user interactions

Previously we used to show four tiles (total interactions, users experiencing failed interactions, users experiencing slow interactions, successful interactions). We are dropping the tile marked “successful interactions”. We are switching the units for the performance tile from number of users facing slow performance to total time spent by users waiting behind the application.

When used to monitor the backend, the Plumbr dashboard will now show the following tiles:

  • Total API calls monitored
  • Number of calls ending in failure
  • Cumulative waiting time for response
Dashboard for a system whose API calls are monitored

New and improved availability summaries

When you look at the summary of availability issues across the applications, you get a denser overview than before. This is because we have introduced
(a) Interactions, sessions, and users in summaries
(b) Charts that include error rates

New availability summary for applications

Appropriate categorization of stuck interactions/API calls

We used to classify “stuck” interactions as failures. The effect of interactions that are stuck and return a response after a long time, are now recorded under performance issues.

Individual Interactions/API calls provided in the context of a specific error/bottleneck

You can now access the list of individual interactions/API calls when investigating from the context of specific errors or bottlenecks. Previously they were available at an application level. We believe that they add more value when exposed contextually.

The general availability of these changes is from 18-Feb-2019. Our Customer success team (csm@plumbr.io) is available to provide any specific assistance or clarify any queries you may have.





ADD COMMENT