CUSTOMER STORY

How Trice Imaging Gets Better Monitoring With Metricly

Featured Commentator:
Kris Kumler, Lead Automation Engineer

The Company

Trice Imaging securely connects medical devices and software to the cloud so that clinicians, patients and organizations can instantaneously access medical images from anywhere and collaborate remotely on cases.

Its award-winning Tricefy service has clinical users in 106 countries including Mount Sinai Hospital in New York, Carnegie Imaging for Women, New York, and the University of Colorado’s Anschutz Medical Campus.

The Problem

When Lead Automation Engineer Kris Kumler started at Trice Imaging more than three years ago, the company was relying on reactive alerts and custom scripts, monitoring their server estate using the open source agent collectd. This setup failed to provide the continuous monitoring Trice aspired to have. Meanwhile, Kumler also needed the capability to act on some of the leading indicators and dynamic thresholds that were based on historical performance for a certain time of day or day of the week.

Dashboard widget with expected values

This graph on a dashboard widget shows the value of a metric along with expected values (in green) based on historic behavior of that metric over time for that time of day and day of the week, and the expected value (purple band) based on its correlation with other interdependent metrics.

The Solution

Kumler looked at several solutions, including Datadog, open source option Graphite, and AWS CloudWatch given that Trice’s environment is hosted in Amazon Web Services (AWS). Ultimately, he signed up for a trial with Metricly, saying that Metricly’s pricing fit within his budget. But during the trial period he noted that Metricly’s customer support “clinched the deal.”

When Kumler first deployed Metricly’s solution, he remembers having some problems with setup. But he was impressed at how his technical salesperson at the time helped him track down the problems. “Some of them were our configuration problems. Other problems led to improved versions of the [Metricly] agent,” Kumler says.

When you’re getting things started up with [a new monitoring solution], it’s more about responding to the right question, instead of brushing us off and having us do it on our own.

Since starting with Metricly’s solution three years ago, Kumler has been impressed at the way Metricly has worked to grow with Trice, as its needs have increased. He adds that Metricly values openness in product decisions and takes customer feedback into account when advancing new features. “Quite a bit of what we suggested was listened to, and we’ve even gotten to see some of the previews of new features before they have come out. It was nice to be able to give that early feedback,” Kumler says.

Green and purple “bands of normalcy” from Metricly’s machine learning

This graph shows Metricly’s behavior learning and bands of normalcy in Trice’s environment. The green band is based on historic analysis while the purple band is determined based on multivariate regression analysis.

The Result

Kumler says that Metricly solved Trice Imaging’s problem by providing continuous monitoring as well as several other features. He says that Metricly’s configurable policies and alerts, all of which may be tied to a baseline for the time of day and element is wonderful for generating meaningful alerts. “This is their big win and fundamental differentiator [and is] much easier than trying to accomplish this otherwise,” Kumler says. He adds that many of the alert policies Trice already has in place operate using the baseline bands of normalcy for automated anomaly detection of performance problems.

Also, Metricly lets him watch when the network is at the lower usage times without hardcoded thresholds that would normally let things slip by, Kumler says. Trice’s traffic varies depending on the time of day.

In addition, Kumler says he has started to explore some of the AWS utilization and cost reports, which provides calculated utilization metrics. Because Trice already uses Metricly for its performance monitoring, the data needed for this type of analysis is already in the platform, making it as easy as flipping a switch to see the capacity and cost analysis reports and recommendations for right-sizing to save money on Trice’s AWS bill.

“I see a lot of promise with that. There is of course, a myriad of services out there that will do that, but it’s not always easy to justify going solely with a service based on that. It can be a good determining factor with us. We’ve got a lot of the right utilization data in Metricly already, so it makes for a very natural place for it to live,” Kumler says.

Metricly’s AWS cost reports show instance level utilization

This graph show the cost of individual instances (left scale) and the utilization of computing resources such as CPU, memory, I/O, network and disk-space on the right-hand scale.

about

Industry: Medical Devices, SaaS
Company Size: 11-50
Headquarters: Del Mar, CA

Objective

To get continuous monitoring with dynamic thresholds & leading indicators from past performance data.

Success

Meaningful alerts and less noise using Metricly machine learning bands of normalcy.

Metricly Anomaly Detection

Look no further.
Start monitoring with Metricly today.

Get Started Free

Look no further.
Start monitoring with Metricly today.

Get Started Free

capterra

Join other DevOps who love Metricly!

Sign up for a free, fully featured, 21-day trial. No credit card required!