d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

MFC r200282, r200290:

Browse Source 
Change gmirror default balance algorithm from "split" to improved "load".
"split" is very ineffective for devices with rotating media as HDDs.
To be effective, it needs that transfer time reduction due to block
splitting was bigger then access time increase due to non-sequential
access. For modern HDDs I was able to reproduce it only with read sizes
of 2MB and above, which is almost not applicable in real life.
"load" algorithm same time is more universal and effective now.
This commit is contained in:
Alexander Motin 2009-12-10 23:51:24 +00:00
parent b6eb79852e
commit 1fea06a081
Notes: svn2git 2020-12-20 02:59:44 +00:00 
svn path=/stable/8/; revision=200373
2 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
sbin/geom/class/mirror/geom_mirror.c
View File

@ -44,7 +44,7 @@ __FBSDID("$FreeBSD$");
uint32_t lib_version = G_LIB_VERSION; uint32_t lib_version = G_LIB_VERSION;
uint32_t version = G_MIRROR_VERSION; uint32_t version = G_MIRROR_VERSION;
static char label_balance[] = "split", configure_balance[] = "none"; static char label_balance[] = "load", configure_balance[] = "none";
static intmax_t label_slice = 4096, configure_slice = -1; static intmax_t label_slice = 4096, configure_slice = -1;
static intmax_t insert_priority = 0, configure_priority = -1; static intmax_t insert_priority = 0, configure_priority = -1;

4
sbin/geom/class/mirror/gmirror.8
View File

@ -24,7 +24,7 @@
.\" .\"
.\" $FreeBSD$ .\" $FreeBSD$
.\" .\"
.Dd August 1, 2009 .Dd December 8, 2009
.Dt GMIRROR 8 .Dt GMIRROR 8
.Os .Os
.Sh NAME .Sh NAME
@ -135,6 +135,7 @@ Specifies balance algorithm to use, one of:
.Bl -tag -width ".Cm round-robin" .Bl -tag -width ".Cm round-robin"
.It Cm load .It Cm load
Read from the component with the lowest load. Read from the component with the lowest load.
This is the default balance algorithm.
.It Cm prefer .It Cm prefer
Read from the component with the biggest priority. Read from the component with the biggest priority.
.It Cm round-robin .It Cm round-robin
@ -142,7 +143,6 @@ Use round-robin algorithm when choosing component to read.
.It Cm split .It Cm split
Split read requests, which are bigger than or equal to slice size on N pieces, Split read requests, which are bigger than or equal to slice size on N pieces,
where N is the number of active components. where N is the number of active components.
This is the default balance algorithm.
.El .El
.It Fl F .It Fl F
Do not synchronize after a power failure or system crash. Do not synchronize after a power failure or system crash.