After SGOS 6.7.x CPU monitor shows CPU 0 is always running higher than the other cores
Last Updated March 07, 2019
After SGOS 6.7.x when navigating under cpu monitor advanced url https://x.x.x.x:8082/Diagnostics/CPU_Monitor/Statistics , it shows CPU 0 is always running higher than the other cores
In SGOS 6.7.x application processing is evenly balanced across all cores compared to 6.6.x and older SGOS versions, where last core(s) were always reserved for TCP/IP (on devices which had 4 or more cores). With new CPU task distribution algorithm implemented in 6.7.x application processing is distributed across all processes (no reserved core for TCP/IP). With this new task distribution algorithm you may observe CPU 0 always has higher utilization compared to other cores.
It is normal for CPU0 (and then CPU1, CPU2, etc) to be loaded higher than the remaining higher numbered CPUs on muti core SG/ASG/SGVA device. This is due to new 6.7.x algorithm used for scheduling works / tasks prefers lower number of cores for coreOS/networking related tasks until it becomes busy, and then starts to use the remaining CPU cores. Essentially, if CPU 0 has available cycles, it will continue to have preferential scheduling causing it to appear to "spike" .
However, since CPU0 is NOT CPU bound there is no impact to end user performance. As CPU 0 continues to increase, some load will fall over to CPU1,then CPU2, CPU3 etc. So you will essentially see a distribution of CPU load, weighted higher for the lower number CPU numbers. But as load increases, it will essentially be evenly distributed when the SG/ASG/SGVA is placed high workload environment or when traffic volume increases.
Subscribing will provide email updates when this Article is updated. Login is required to Subscribe