trapexit / mergerfs

a featureful union filesystem
http://spawn.link
Other
4.03k stars 168 forks source link

Benchmarking tool #1331

Closed oregonpillow closed 2 months ago

oregonpillow commented 2 months ago

Is your feature request related to a problem? Please describe. With the various options available, it seems the only way to find out which settings are optimal for performance are to simply try different configurations/combinations.

Describe the solution you'd like Would a benchmarking tool make sense for the project? Something that would automate some of the dd benchmark tests for different mergerfs options (also clear kernel caches between tests) and return a summary of results for different configurations.

Happy to try and contribute something if you think it has value.

trapexit commented 2 months ago

I would say "no". There are many options but they aren't performance options. They are functional options. They enable or disable specific, sometimes completely incompatible options. Doing a benchmark would not be meaningful unless you absolutely didn't care at all about the function of the product... which likely no one would.

oregonpillow commented 2 months ago

I understand, despite the options being functional I had the impression that for some options, it was less than obvious for a given system, which options should be enabled for better performance. Without reading up on all of the options so far, I wrongly assumed that some had minimal effect on functionality.

Unfortunately, given the number of variables it has been difficult to find a single set of settings which provide optimal performance

But no worries just thought i'd ask. Thanks for quick reply

trapexit commented 2 months ago

The options are not performance related per se. They may affect performance but if it was merely a "make faster" switch it would just be enabled by default. Hence the explicit "NOTE: be sure to read about these features before changing them to understand what behaviors it may impact"