How do I characterize my usecase so that I can simulate it with fio the closest?

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hello all,

The company I work with uses USB Sticks as a crucial part of the
product and I was assigned a task to bring up a pragmatic method to
characterize USB Sticks from different models/vendors/manufacturers.

I intend to write a simple pytest based framework that will use fio to
characterize the USB sticks systematically.
This would allow me to perform hundreds of tests in sampled devices to
either approve or reprove that model/vendor/manufacturer, based on the
performance parameters.

To start with, I plan to perform a series of synthetic tests - where I
don't intend to simulate the real use case, but rather, just to
compare different device's performances against each other in a
handful of made-up stress scenarios.

At a second moment, tough, I would in fact like to implement a test
that is the closest as possible to the real process we have during USB
stick flashing

Question:
How do I characterize my use case so that I can simulate it with fio?

My process is something like "./proprietary_script.sh --device
/dev/sdc", where "proprietary_script.sh" is a single threaded app that
mounts the USB Stick and performs a series of copies (with cp) from
the host's SSD to the USB Stick fs.

How can I know what are the values of IO type, IO engine, IO size,
block size, etc to call fio with, so that fio will simulate my process
the closest? How can I gather such information?

By the way, the OS is a Ubuntu 14.

Thank you!

Giuliano Galhardi Motter
mottergiuliano@xxxxxxxxx




[Index of Archives]     [Linux Kernel]     [Linux SCSI]     [Linux IDE]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux SCSI]

  Powered by Linux