Hardware monitoring across instances and Docker containers

We are releasing a new view today - Hardware.

The aim of this view is to allow you to:

  • Monitor usage of physical resources by your JVMs across the cluster
  • Notice how changes to your hardware (scaling up/down) affect your application performance
  • Allow optimization of hardware resources to best match your application’s requirements, so you are not over/under-spending on hardware

This is specially useful for people who use autoscaling or run inside containers like Docker to dynamically alter their physical resources on-demand.

You can see this view in your dashboard by going to Overview > Hardware.

Lets walk through each of the panels. We will use 2 JVMs A & B to show example data.

JVM Count

This shows you the number of JVMs active for this application

Core Count

This shows the total number of CPU Cores available to this application across JVMs.

Eg - A is deployed on a 4 core box, B on a 2 core box.
This graph will show 4+2 = 6 cores available to your app.

Total CPU

Using this graph along with the ‘Core Count’ graph, you can optimize the number of CPU cores available to your application.

It shows the total CPU used across your JVMs. 

Eg - A is using 50% of total CPU and B is using 40%.
This graph will show (50+40)/2 = 45% CPU usage.

Total Memory

This graph allows you to optimize the memory available to your applications and tune the heap to match your needs.

It shows the total RAM, Total Available Heap and Total Used Heap.

A has 4GB RAM with Max Heap of 3GB and using 2GB.
B has 2GB RAM with Max Heap of 1GB and using 0.5GB.
The graph will show:
Total RAM = 4+2 = 6GB, Total Available Heap = 3+1 = 4GB, Total Used Heap = 2+0.5 = 2.5GB

Longest GC Pause

Shows the single longest GC Pause across your JVMs for a time period.

Eg, At time X, A has the longest GC pause of 10ms, and B of 50ms.
This graph will show  50ms.

Total Thread Count

Shows the total thread count for your application across JVMs.

Eg, A has 100 threads, B has 200 threads.
This graph will show 100 + 200 = 300 threads.

Next Steps

  • Make sure you have atleast Agent 6.1.20 to see all the data in this view.
  • See this view now on our Live Demo.
Show Comments