From: Philipp Reisner <philipp.reisner@linbit.com>
To: drbd-dev@lists.linbit.com
Subject: [Drbd-dev] internal vs external meta data [reminder]
Date: Tue, 31 Jan 2006 19:00:15 +0100 [thread overview]
Message-ID: <200601311900.15806.philipp.reisner@linbit.com> (raw)
internal vs external meta data
Theoretical it should be quicker in case you use external meta
data. Because AL-misses cause the disk-head to seek to the meta-
data area, write to the AL, and then seek back to the data location.
Real world measurement shows that the opposite is currently the case:
internal meta data
257 misses (outer sectors)
19.62 MB/sec (1077936128 B / 00:52.390089)
19.28 MB/sec (1077936128 B / 00:53.314101)
19.31 MB/sec (1077936128 B / 00:53.229198)
0 misses (outer sectors)
20.72 MB/sec (1077936128 B / 00:49.611017)
20.95 MB/sec (1077936128 B / 00:49.074782)
20.77 MB/sec (1077936128 B / 00:49.502596)
external meta data:
257 misses (outer sectors)
16.40 MB/sec (1077936128 B / 01:02.677786)
16.32 MB/sec (1077936128 B / 01:02.994084)
16.51 MB/sec (1077936128 B / 01:02.271167)
0 misses (outer sectors)
16.46 MB/sec (1077936128 B / 01:02.453234)
16.51 MB/sec (1077936128 B / 01:02.261244)
16.69 MB/sec (1077936128 B / 01:01.606018)
With the write to the meta-data the disk queue runs more frequent,
in case the meta-data is internal, this causes the higher performance
because in the same disk queue are also the blocks for the data
area.
Completely unclear is why the two test sets with no misses completed
with different results. The only difference was the location of
the meta data, but there are now writes to the meta-data area,
if there are no misses.
-philipp
--
: Dipl-Ing Philipp Reisner Tel +43-1-8178292-50 :
: LINBIT Information Technologies GmbH Fax +43-1-8178292-82 :
: Schönbrunnerstr 244, 1120 Vienna, Austria http://www.linbit.com :
reply other threads:[~2006-01-31 18:00 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=200601311900.15806.philipp.reisner@linbit.com \
--to=philipp.reisner@linbit.com \
--cc=drbd-dev@lists.linbit.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox