Standardize data rate units in IO statistics

[ttys42]

Is your feature request related to a problem? Please describe.
Currently the beegfs stats server command displays data rates using mixed dynamic units (e.g., jumping between KB, MB, and TB based on the instantaneous throughput). This makes it difficult for users to visually compare data rates at a glance or monitor trends effectively

Describe the solution you'd like
Display data rates in a constant unit. Can be either hardcoded or maybe toggled with a flag --display-unit. Current behaviour could be saved by --display-unit dynamic

Describe alternatives you've considered

Additional context

[iamjoemccormick]

Hi @ttys42 , this is a good idea. There is actually a --raw global flag which already achieves this, however it always outputs in bytes which is less human readable.

I would propose to keep the current behavior the default (e.g., --display-unit dynamic) to avoid surprising anyone with a behavior change, but I'm open to defaulting to MiB or GiB if thats what people are going to want 80%+ of the time.

[ttys42]

That --raw flag is quite handy! Combined with a bit of awk, I can translate to alternate units (but formatting is a bit off)
While both --raw and the current default behavior are useful and should definitely be kept, neither is ideal for quick visual parsing.
Adding an alternate option to automatically format outputs in MiB or GiB would be a good improvement.

[ttys42]

Side note:

Consider switching the data columns from left-aligned to right-aligned. This will drastically improve overall readability. As you can see with the SENT column, the values are very easy to compare because the 'MiB' strings line up perfectly. In contrast, left-aligned columns like READ are much harder to scan visually

More drastically with --raw