Technical Documentation

Understanding Overall System Condition and Fabric Load

You can view the overall Junos Space system condition and fabric load from the platform application dashboard or from the Administration workspace landing page.

System Condition

To calculate the overall system condition, Junos Space uses an algorithm based on cluster health and node-function health:

  • Cluster health indicates the percentage of nodes in the fabric that are currently running.

    For example, if only three nodes are reachable in a four-node fabric, cluster health is 75%.

  • Load-balancer health indicates the percentage of nodes (enabled for load balancing) that are running the load balancing process.

    For example, if two nodes are enabled for load balancing and the load-balancing process is running on only one node, the load-balancing health is 50%.

  • Database health indicates the percentage of nodes (enabled for database requests) that are running the database process.

    For example, if two nodes are enabled as database server and the database process is running on only one node, then database health is 50%.

  • Application-logic health indicates the percentage of nodes (enabled for application logic (DML and business logic)) that are running the application-logic process.

    For example, if three nodes are enabled for application logic and the application-logic process is running on only two nodes, then application-logic health is 67%.

Junos Space retrieves data on the nodes and the node functions running, and then applies the following algorithm to determine the overall system condition:

overall system condition = [ (number of nodes running) / (number of nodes in fabric) ] * [ (number of nodes running load balancing process) / (number of nodes enabled for load balancing) ] * [ (number of nodes running database server process) / (number of nodes enabled as database server) ] * [ (number of nodes running application logic process) / (number of nodes enabled for application logic) ]

Using the preceding examples for cluster health and node-function health, the overall system condition is expressed as a percentage:

overall system condition = 75% * 50% * 50% * 67% = 12.5%

The Overall System Condition window indicates Poor (0–30%), Average (30–70%), or Good (70–100%), based on the value the algorithm returns.

Image s600450.gif

The overall system health indicates 0% (Poor) when any one of the following conditions is detected:

  • No nodes in the fabric are running.
  • No nodes enabled for load balancing are running the load balancing process.
  • No nodes enabled for database requests are running the database process.
  • No nodes enabled for application logic are running the application logic process.

Fabric Load

The Fabric Load chart displays the average CPU usage across all nodes that are running in the fabric.

Image s600451.gif

Junos Space uses the following algorithm to determine the fabric load:

fabric load = [ total CPU usage for all nodes running ] / [number of nodes running ]

For example, given a fabric with three nodes running and CPU usage of 80%, 30%, and 10%, respectively, the fabric load is 40%. The following example illustrates how the fabric load is calculated.

fabric load = [ 80% + 30% +10% ] / 3 fabric load = 120% / 3 fabric load = 40%

To view the average CPU use at a specific data point, drag the mouse over the data point of interest.

To obtain details about the status of the fabric, click any data point in the graph. The Fabric Monitoring window is displayed and shows detailed status for each node in the fabric. Status information includes CPU, disk, and memory usage and indicates up or down status for each node function enabled on the node.

Published: 2009-12-16