CPU data, control and analytics plane utilization

Question

Hi everyone,&nbsp;Wondering if there is any "quick" way of extracting the CPU statistics for Data, Control and Analytics plane utilization via iControl ?As far as I read, &nbsp;Even-numbered logical cores (hyperthreads) are allocated to TMM, while odd numbered cores are available for other processes, while last core is used for analytics. &nbsp;Do I need to do the math myself ?

welinton_trigueiro · Answer

&nbsp;When we execute a GET request to https://&lt;BIG-IP-ADDRESS&gt;/mgmt/tm/sys/host-info, the BIG-IP returns the information in JSON format.I had created a Python script that wasn't well formatted, but after adjusting a few things using AI, it seems to be working well now.In the case of the .py script, I am separating even and odd CPUs in an attempt to visualize the data plane (even CPU) and control plane (odd CPU).&nbsp;&nbsp;If it helps, here is the script.

robert_f5 · Answer

Hi,&nbsp;That's what I'm looking for as well. Have you found a solution?&nbsp;Regards

nikoolayy1 · Answer

For BD utilization I used this API but you can modify for your needs.
&nbsp;
mgmt/tm/analytics/asm-cpu/report?options=view-by,virtual,measures,%7Bbd-cpu-utlization%7D
&nbsp;
APIRef_tm_analytics

robert_f5 · Answer

Hi,Thanks for the answer. Given the figure below, the average of odd-numbered CPUs represents the orange line (Control Plane), and the average of even-numbered CPUs represents the Data Plane, am I correct? What about Analysis Plane and System Average?By the way, could you share the script or provide a link to access it?Regards

welinton_trigueiro · Answer

&nbsp;Each CPU core is designated as data, control, or analytics, with all usage assigned to the core's designated category. Each even-numbered core is designated as a data plane, each odd-numbered core as a control plane, and the highest odd-numbered core as an analytics plane.Analytics plane-related processes are typically pinned to the highest odd-numbered core, and no control or analytics processes have access to the even-numbered cores. The processes pinned to the analytics plane usually consume all available CPU cycles when running, which is why they are assigned to a single specific core.This division occurs when the CPU supports Hyper-Threading Technology (HT Technology). Overview of the HTSplit feature -&nbsp;https://my.f5.com/manage/s/article/K23505424&nbsp;The BIG-IP VE system does not support this feature on generic hypervisors such as VMware ESX/i, where each vCPU will be allocated one TMM process. &nbsp;https://my.f5.com/manage/s/article/K15468#tmsh&nbsp;If you are seeing control plane and data plane CPU statistics in the dashboard, this is related to bug I id 969329 - https://my.f5.com/manage/s/article/K000132982 &nbsp;here is the script:import requests
from requests.auth import HTTPBasicAuth
from tabulate import tabulate
import json

# BIG-IP
BIGIP_ADDRESS = "192.168.1.10"
USERNAME = "admin"
PASSWORD = "admin"

# disable warnings de SSL
requests.packages.urllib3.disable_warnings()

def get_cpu_stats():
    url = f"https://{BIGIP_ADDRESS}/mgmt/tm/sys/host-info"
    response = requests.get(url, auth=HTTPBasicAuth(USERNAME, PASSWORD), verify=False)
    
    if response.status_code == 200:
        data = response.json()

cpu_info_path = data.get("entries", {}).get("https://localhost/mgmt/tm/sys/host-info/0", {}).get("nestedStats", {}).get("entries", {}).get("https://localhost/mgmt/tm/sys/hostInfo/0/cpuInfo", {}).get("nestedStats", {}).get("entries", {})
        
        odd_cpus = []
        even_cpus = []
        
        if cpu_info_path:
            for cpu_key, stats in cpu_info_path.items():
                cpu_id = stats["nestedStats"]["entries"].get("cpuId", {}).get("value")
                one_min_usage = stats["nestedStats"]["entries"].get("oneMinAvgUser", {}).get("value", "N/A")
                five_min_usage = stats["nestedStats"]["entries"].get("fiveMinAvgUser", {}).get("value", "N/A")
                
                if cpu_id % 2 == 0:
                    even_cpus.append([f"cpu{cpu_id}", one_min_usage, five_min_usage])
                else:
                    odd_cpus.append([f"cpu{cpu_id}", one_min_usage, five_min_usage])

print("
Table of Even CPUs:")
            print(tabulate(even_cpus, headers=["CPU", "1 Min Usage (%)", "5 Min Usage (%)"], tablefmt="pretty"))
            
            print("
Table of Odd CPUs:")
            print(tabulate(odd_cpus, headers=["CPU", "1 Min Usage (%)", "5 Min Usage (%)"], tablefmt="pretty"))
        
        else:
            print("Unable to find CPU information in the returned JSON.")
    else:
        print(f"Failed to retrieve data: {response.status_code} - {response.text}")

get_cpu_stats()&nbsp;

Forum Discussion

CPU data, control and analytics plane utilization

Recent Discussions

Should config via cli rather than gui?

APM Modern Customization - modify Header in user-common.js and form in user-logon.js

F5 Health Monitor Receive String as JSON

Cloudflare True-Client-IP as Persistence

Get actual client ips in splunk

Related Content

JWT authorization with NGINX Ingress Controller

F5 Analytics iApp

Distributed caching of authentication requests with NGINX Ingress Controller

Introducing TCP Analytics

Using BIG-IP Kubernetes Gateway API Controller

ABOUT DEVCENTRAL

RESOURCES

SUPPORT

PARTNERS