1 =========
2 dm-switch
3 =========
4
5 The device-mapper switch target creates a device that supports an
6 arbitrary mapping of fixed-size regions of I/O across a fixed set of
7 paths. The path used for any specific region can be switched
8 dynamically by sending the target a message.
9
10 It maps I/O to underlying block devices efficiently when there is a large
11 number of fixed-sized address regions but there is no simple pattern
12 that would allow for a compact representation of the mapping such as
13 dm-stripe.
14
15 Background
16 ----------
17
18 Dell EqualLogic and some other iSCSI storage arrays use a distributed
19 frameless architecture. In this architecture, the storage group
20 consists of a number of distinct storage arrays ("members") each having
21 independent controllers, disk storage and network adapters. When a LUN
22 is created it is spread across multiple members. The details of the
23 spreading are hidden from initiators connected to this storage system.
24 The storage group exposes a single target discovery portal, no matter
25 how many members are being used. When iSCSI sessions are created, each
26 session is connected to an eth port on a single member. Data to a LUN
27 can be sent on any iSCSI session, and if the blocks being accessed are
28 stored on another member the I/O will be forwarded as required. This
29 forwarding is invisible to the initiator. The storage layout is also
30 dynamic, and the blocks stored on disk may be moved from member to
31 member as needed to balance the load.
32
33 This architecture simplifies the management and configuration of both
34 the storage group and initiators. In a multipathing configuration, it
35 is possible to set up multiple iSCSI sessions to use multiple network
36 interfaces on both the host and target to take advantage of the
37 increased network bandwidth. An initiator could use a simple round
38 robin algorithm to send I/O across all paths and let the storage array
39 members forward it as necessary, but there is a performance advantage to
40 sending data directly to the correct member.
41
42 A device-mapper table already lets you map different regions of a
43 device onto different targets. However in this architecture the LUN is
44 spread with an address region size on the order of 10s of MBs, which
45 means the resulting table could have more than a million entries and
46 consume far too much memory.
47
48 Using this device-mapper switch target we can now build a two-layer
49 device hierarchy:
50
51 Upper Tier - Determine which array member the I/O should be sent to.
52 Lower Tier - Load balance amongst paths to a particular member.
53
54 The lower tier consists of a single dm multipath device for each member.
55 Each of these multipath devices contains the set of paths directly to
56 the array member in one priority group, and leverages existing path
57 selectors to load balance amongst these paths. We also build a
58 non-preferred priority group containing paths to other array members for
59 failover reasons.
60
61 The upper tier consists of a single dm-switch device. This device uses
62 a bitmap to look up the location of the I/O and choose the appropriate
63 lower tier device to route the I/O. By using a bitmap we are able to
64 use 4 bits for each address range in a 16 member group (which is very
65 large for us). This is a much denser representation than the dm table
66 b-tree can achieve.
67
68 Construction Parameters
69 =======================
70
71 <num_paths> <region_size> <num_optional_args> [<optional_args>...] [<dev_path> <offset>]+
72 <num_paths>
73 The number of paths across which to distribute the I/O.
74
75 <region_size>
76 The number of 512-byte sectors in a region. Each region can be redirected
77 to any of the available paths.
78
79 <num_optional_args>
80 The number of optional arguments. Currently, no optional arguments
81 are supported and so this must be zero.
82
83 <dev_path>
84 The block device that represents a specific path to the device.
85
86 <offset>
87 The offset of the start of data on the specific <dev_path> (in units
88 of 512-byte sectors). This number is added to the sector number when
89 forwarding the request to the specific path. Typically it is zero.
90
91 Messages
92 ========
93
94 set_region_mappings <index>:<path_nr> [<index>]:<path_nr> [<index>]:<path_nr>...
95
96 Modify the region table by specifying which regions are redirected to
97 which paths.
98
99 <index>
100 The region number (region size was specified in constructor parameters).
101 If index is omitted, the next region (previous index + 1) is used.
102 Expressed in hexadecimal (WITHOUT any prefix like 0x).
103
104 <path_nr>
105 The path number in the range 0 ... (<num_paths> - 1).
106 Expressed in hexadecimal (WITHOUT any prefix like 0x).
107
108 R<n>,<m>
109 This parameter allows repetitive patterns to be loaded quickly. <n> and <m>
110 are hexadecimal numbers. The last <n> mappings are repeated in the next <m>
111 slots.
112
113 Status
114 ======
115
116 No status line is reported.
117
118 Example
119 =======
120
121 Assume that you have volumes vg1/switch0 vg1/switch1 vg1/switch2 with
122 the same size.
123
124 Create a switch device with 64kB region size::
125
126 dmsetup create switch --table "0 `blockdev --getsz /dev/vg1/switch0`
127 switch 3 128 0 /dev/vg1/switch0 0 /dev/vg1/switch1 0 /dev/vg1/switch2 0"
128
129 Set mappings for the first 7 entries to point to devices switch0, switch1,
130 switch2, switch0, switch1, switch2, switch1::
131
132 dmsetup message switch 0 set_region_mappings 0:0 :1 :2 :0 :1 :2 :1
133
134 Set repetitive mapping. This command::
135
136 dmsetup message switch 0 set_region_mappings 1000:1 :2 R2,10
137
138 is equivalent to::
139
140 dmsetup message switch 0 set_region_mappings 1000:1 :2 :1 :2 :1 :2 :1 :2 \
141 :1 :2 :1 :2 :1 :2 :1 :2 :1 :2
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.