brintos

brintos / linux-shallow public Read only

0
0
Text · 3.2 KiB · facf277 Raw
102 lines · plain
1===============2dm-service-time3===============4 5dm-service-time is a path selector module for device-mapper targets,6which selects a path with the shortest estimated service time for7the incoming I/O.8 9The service time for each path is estimated by dividing the total size10of in-flight I/Os on a path with the performance value of the path.11The performance value is a relative throughput value among all paths12in a path-group, and it can be specified as a table argument.13 14The path selector name is 'service-time'.15 16Table parameters for each path:17 18    [<repeat_count> [<relative_throughput>]]19	<repeat_count>:20			The number of I/Os to dispatch using the selected21			path before switching to the next path.22			If not given, internal default is used.  To check23			the default value, see the activated table.24	<relative_throughput>:25			The relative throughput value of the path26			among all paths in the path-group.27			The valid range is 0-100.28			If not given, minimum value '1' is used.29			If '0' is given, the path isn't selected while30			other paths having a positive value are available.31 32Status for each path:33 34    <status> <fail-count> <in-flight-size> <relative_throughput>35	<status>:36		'A' if the path is active, 'F' if the path is failed.37	<fail-count>:38		The number of path failures.39	<in-flight-size>:40		The size of in-flight I/Os on the path.41	<relative_throughput>:42		The relative throughput value of the path43		among all paths in the path-group.44 45 46Algorithm47=========48 49dm-service-time adds the I/O size to 'in-flight-size' when the I/O is50dispatched and subtracts when completed.51Basically, dm-service-time selects a path having minimum service time52which is calculated by::53 54	('in-flight-size' + 'size-of-incoming-io') / 'relative_throughput'55 56However, some optimizations below are used to reduce the calculation57as much as possible.58 59	1. If the paths have the same 'relative_throughput', skip60	   the division and just compare the 'in-flight-size'.61 62	2. If the paths have the same 'in-flight-size', skip the division63	   and just compare the 'relative_throughput'.64 65	3. If some paths have non-zero 'relative_throughput' and others66	   have zero 'relative_throughput', ignore those paths with zero67	   'relative_throughput'.68 69If such optimizations can't be applied, calculate service time, and70compare service time.71If calculated service time is equal, the path having maximum72'relative_throughput' may be better.  So compare 'relative_throughput'73then.74 75 76Examples77========78In case that 2 paths (sda and sdb) are used with repeat_count == 12879and sda has an average throughput 1GB/s and sdb has 4GB/s,80'relative_throughput' value may be '1' for sda and '4' for sdb::81 82  # echo "0 10 multipath 0 0 1 1 service-time 0 2 2 8:0 128 1 8:16 128 4" \83    dmsetup create test84  #85  # dmsetup table86  test: 0 10 multipath 0 0 1 1 service-time 0 2 2 8:0 128 1 8:16 128 487  #88  # dmsetup status89  test: 0 10 multipath 2 0 0 0 1 1 E 0 2 2 8:0 A 0 0 1 8:16 A 0 0 490 91 92Or '2' for sda and '8' for sdb would be also true::93 94  # echo "0 10 multipath 0 0 1 1 service-time 0 2 2 8:0 128 2 8:16 128 8" \95    dmsetup create test96  #97  # dmsetup table98  test: 0 10 multipath 0 0 1 1 service-time 0 2 2 8:0 128 2 8:16 128 899  #100  # dmsetup status101  test: 0 10 multipath 2 0 0 0 1 1 E 0 2 2 8:0 A 0 0 2 8:16 A 0 0 8102