Skip to content

Latest commit

 

History

History
104 lines (73 loc) · 2.87 KB

File metadata and controls

104 lines (73 loc) · 2.87 KB

Comparison with alternative tools

Missing anything? File an issue.

Atop

https://www.atoptool.nl/

  • Terminal interface
    • Supports sorting, scrolling
  • In-depth host-level stats
  • Built in support for persisting/replaying historical data
  • atopsar for scriptable access to historical data

Drawbacks

  • No cgroup awareness
  • May suffer priority inversions while host is under resource contention
  • On-disk data has custom delta compression and may be corrupted easily

cAdvisor

https://github.com/google/cadvisor

  • Web interface
  • REST API for access to current data
  • Tight integration with containers and container runtimes (eg docker)
  • Reports high level stats on containers
  • Plugin architecture for persisting historical data

Drawbacks

  • Requires external storage service to persist historical data
  • cgroup1 only
  • Limited support for host level stats
  • No built in terminal interface

htop

https://htop.dev/

  • Wonderful terminal interface
    • Supports searching, filtering, sorting, scrolling
  • Process tree view
  • Supports system actions (killing processes, straceing processes)

Drawbacks

  • No support for persisting/replaying historical data
  • No cgroup awareness
  • Limited host-level stats

sysstat (sar)

https://github.com/sysstat/sysstat

  • CLI-based tools like sar, iostat, pidstat, and others
  • Collects a wide range of system performance metrics (CPU, memory, IO, etc.)
  • Mature and widely used for historical performance tracking
  • Data can be collected periodically via cron or systemd timers
  • Data stored in a binary format, accessible through CLI

Drawbacks

  • No interactive or web interface
  • Lacks container or cgroup-specific visibility
  • Requires parsing and scripting to extract structured data
  • Binary logs are not human-readable without sar

collectd

https://collectd.org/

  • Daemon-based metrics collector
  • Plugin architecture for system and application metrics (disk, memory, network, etc.)
  • Can push metrics to various backends (Graphite, InfluxDB, Prometheus via exporter)
  • Lightweight and extensible with community plugins
  • Suitable for long-term storage and visualization setups

Drawbacks

  • No built-in UI or terminal interface
  • Configuration can be complex
  • Requires external tooling for data visualization
  • No direct cgroup or container integration by default

below

https://github.com/facebookincubator/below

  • Terminal interface
    • Supports filtering, zooming, pausing, sorting, scrolling
  • Built in support for persisting/replaying historical data
  • In-depth host-level stats
  • cgroup awareness with cgroup tree view
  • below dump for scriptable access to historical data
  • Goes to great lengths to avoid priority inversions during host resource contention

Drawbacks

  • cgroup2 only
  • No built in data compression (recommends btrfs transparent compression)