Performance Counters for the Linux Operating System

/ Performance engineering / By

The Linux operating system is a well-liked option for servers. It is a highly dependable and secure operating system that is open-source and free. Additionally, it is simple to customize and works with the majority of hardware and applications. Numerous functions, including web hosting, virtualization, file and print services, and database management, are performed on Linux servers. Applications for the Internet of Things (IoT), big data analytics, and cloud computing all employ Linux servers. Linux servers may run on a variety of hardware and provide scalability and flexibility. They are also renowned for their performance, security, and stability. Linux servers are affordable, making them a preferred option for companies of all sizes.

On the Linux operating system, some commands can help to get information on resource usage and details.

CPU

  1. lscpu
  2. top
  3. mpstat
  4. vmstat
  5. SAR
  6. Htop
  7. Neon

Memory

  1. top
  2. free
  3. dmidecode
  4. /proc/meminfo is less

Disk

  1. sysstat
  2. iostat
  3. Performance Counters for the Linux OS

System

  1. CPU User: The amount of time spent executing non-kernel programs (including enjoyable time). time spent on the CPU as a percentage.
  2. CPU System: The amount of time used to execute kernel code. time spent on the CPU as a percentage.
  3. CPU Idle: Idle time on the CPU. Before Linux 2.5.41, IO-wait time was included in this. time spent on the CPU as a percentage.
  4. Interruptions Count: The number of interruptions every second, including the clock, is known as the interruptions count.
  5. Context Switch Rate: The frequency of context transitions.
  6. Context Switch Count: The number of context switches made by each CPU each second.
  7. Processes Runnable: The number of processes that are awaiting execution.
  8. Processes Runnable per CPU: The quantity of processes each CPU has available for execution.
  9. Processes Sleeping: The number of processes resting soundly.
  10. Processes Swapped: The number of running processes that have been switched out. Although this field is calculated, Linux never uses desperate swaps.
  11. Swap in: Amount of RAM that was swapped in from disk (/s)
  12. Swap out: Memory transferred to disc in the specified amount of time (/s)

Memory

  1. Whole Memory: The whole amount of RAM that is usable (i.e., physical RAM less a few reserved bits and the kernel binary code).
  2. Memory Used: Total memory used minus free memory.
  3. Memory Free: The total (overall stat) of LowFree and HighFree.
  4. Memory Buffered: Storage of memory in a buffer cache. metric today is largely meaningless.
  5. Memory Cache: Memory in the disc cache, excluding SwapCache, is referred to as the memory cache.
  6. User Memory: Memory consumed by the user, excluding caches and buffers. Memory Total – Memory Free – Memory Buffered – Memory Cache = User Memory.
  7. % User Memory: The proportion of memory that is being used (% User Memory = User Memory / Total Memory * 100)
  8. Total Swap: Total physical swap memory capacity.
  9. Swap Free: Total free swap memory is referred to as swap-free.
  10. Swap Total – Swap Free: Swap Used.

Disk (per device)

  1. Disc Read Request Merged: The number of read requests that are combined and sent to the device every second.
  2. Disc Write Request Merged: The quantity of merged write requests sent to the device each second.
  3. Disc Read The rate at which the device receives read requests.
  4. Disc Write The rate at which the device receives requests to write data.
  5. Disk Read (sector): The number of sectors reads from the device per second is known as the disc read (sector).
  6. Disc Write (sector): The rate at which a device is written using sectors.
  7. Disk Read (KB): The number of kilobytes reads from the device per second is known as the disc read (KB).
  8. Disc Write (KB): The amount of data written to the drive in kilobytes each second.
  9. Disc Request Size: The typical size of requests sent to the device, measured in sectors.
  10. Queue Length: The typical length of the queue for requests sent to the device.
  11. IO Wait The typical amount of time (in milliseconds) required for an I/O request to be fulfilled by the device. This comprises the amount of time requests are held in a queue and the time it takes to process them.
  12. Disc Service Time: The typical response time (measured in milliseconds) for requests for I/O made of the hardware.
  13. IO CPU Time: The portion of CPU time used to send I/O requests to the device (the device’s bandwidth utilization). When this number approaches 100%, device saturation occurs.

Processes (per process)

Each process has access to the counters in the processes section. Select the process identification in the counter definition window outside of the wizard. The selection button opens a picker for the process identification.

  1. Process Memory Usage: Resident size (kb) in terms of process memory usage. the physical memory that a task hasn’t swapped out.
  2. Process CPU Time: The task share of the Total CPU time spent on CPU tasks since the last update.

Network (per interface)

The network section counters are accessible by the interface. Select the interface name in the counter definition box outside of the wizard. The Fill button accesses a network interface selector.

  1. Incoming Bytes/s: The amount of data received by the network interface per second in terms of incoming bytes/s.
  2. Incoming packets: The number of packets that the network interface receives every second is known as incoming packets.
  3. incoming packet errors/s: The number of bad packets received by the network interface per second in terms of incoming packet errors/s.
  4. Incoming Packets Dropped/s: Number of Incoming Packets Dropped/s: The number of packets that the network interface drops per second.
  5. Incoming Packet Error %: The percentage of broken packets received by the network interface is known as the “Incoming Packet Error %” formula (Incoming Packet Error % = Incoming Packet Errors / Incoming Packets *100 ).
  6. Outgoing Bytes/s – The number of bytes that the network interface sends every second is known as Outgoing Bytes/s.
  7. Outgoing packets/s: The number of packets that the network interface sends every second is known as outgoing packets/s.
  8. Outgoing Packet Errors/s: Outgoing Packet Errors/s: The rate at which the network interface generates bad packets.
  9. Outgoing Packets Dropped/s: Number of outgoing packets sent by the network interface that are lost every second.
  10. Outgoing Packet Errors: The percentage of packets sent across the network interface that contain errors is known as “Outgoing Packet Error %” (Outgoing Packet Error % = Outgoing Packet Errors/Outgoing Packets *100).
  11. Packet Collisions: The number of packet collisions identified by the network interface.

TCP

  1. Incoming Segments: Segments Arriving Per Second The number of TCP segments arriving each second.
  2. Outgoing segments/s: The number of TCP segments sent every second is known as outgoing segments/s.
  3. Retransmitted segments/s: Number of TCP segments retransmitted every second, measured in retransmitted segments/s.
  4. % Rebroadcast Segments: Rebroadcast Segments / Outgoing Segments *100 = % Rebroadcast Segments, which is the percentage of rebroadcast segments.
Scroll to Top