linux-kernel - Re: Big I/O latencies, except when iotop is hooked

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <201205092043.11544.arnd@arndb.de>
Date:	Wed, 9 May 2012 20:43:11 +0000
From:	Arnd Bergmann <arnd@...db.de>
To:	Felipe Contreras <felipe.contreras@...il.com>
Cc:	Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: Big I/O latencies, except when iotop is hooked

On Wednesday 09 May 2012, Felipe Contreras wrote:
> And it doesn't tell me why I see the issue on my SSD as well; albeit
> with much less delay.
> 
> I'll try what you suggest, maybe I will get better performance by
> aligning the partitions, but the real issue seems to lie elsewhere,
> right?

For anything but the absolute worst devices, misaligning the partition
should cause a linear increase in latency (maybe factor 2 in the worst
case), so it doesn't explain the entire thing. However, you can use
flashbench to find out the erase block size. If it's larger than 4MB,
you definitely have a problem even with ext4 or btrfs. The latencies
you get from garbage collection for a single write are roughly the time
it takes to write one erase block, and that can be up to a second in the
worst case (slow media with large erase blocks).

	Arnd
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/