2 Planning LSM Volumes and Disk Groups

You must plan your LSM configuration before you can use LSM volumes for applications or file systems. Planning your LSM configuration includes deciding:

How many volumes you want, and the number and type of data plexes the volumes will use

How many disk groups you need, and which disks you will configure in a disk group

This chapter provides information and worksheets to assist you in planning LSM volumes and disk groups. You might want to make copies of the blank worksheets for future use.

2.1 Planning LSM Volumes

Planning LSM volumes includes deciding what attributes you want the LSM volumes to have. An LSM volume has two types of attributes:

Attributes for which you must provide a value, as described in Table 2-1.

Attributes that are assigned a default value, which you can change, as described in Table 2-2.

Table 2-1: LSM Volume Attributes with No Default Values

Attribute

Notes

Volume name

Can be 31 alphanumeric characters but cannot include a space or slash (/).

Must be unique in the disk group where you create the volume.

Volume size or length

The amount of space needed for the data in the LSM volume.

You can specify volume size in sectors (the default), kilobytes, megabytes, gigabytes, or terabytes.

Table 2-2: LSM Volume Attributes with Default Values

Attribute	Notes and Default Value
Number of plexes	LSM volumes can have up to 32 plexes, with the following restrictions or recommendations: Volumes that use a RAID 5 plex have only one data plex (RAID 5 plexes cannot be mirrored) and can have up to 31 log plexes. Volumes that use concatenated or striped plexes can have any combination of data and log plexes for a total of 32. At least one plex should be a log plex. Default: One concatenated data plex, no log plex.
Log plex size	For volumes less than or equal to 1 GB that use mirror plexes, the default DRL is 65 blocks to allow for migration to a TruCluster environment. The minimum DRL size is approximately 2 blocks per GB of volume size. (You can use the minimum if you know the LSM configuration will not be used in a cluster.) For volumes that use a RAID 5 plex, the log plex size is [10 * (number of columns * data unit size)].
Plex type	A plex type is either concatenated, striped, or RAID 5. You can mirror concatenated or striped plexes. Default: Concatenated, no mirror. See Table 2-3 for information on choosing a plex type.
Name of the disk group where you will create the volume	A volume can be in only one disk group. Default: rootdg disk group.
LSM disks that the volume will use	If the volume has a striped or RAID 5 plex, each column must be of equal size and be on different disks, preferably on different buses. If the volume has mirror plexes, each data plex should use disks on different buses, and the DRL plex should be on a disk that is not used for a data plex. Default: LSM chooses the disks.
Usage type of the volume	Use `fsgen` for volumes that use concatenated or striped plexes and contain a file system. Use `gen` for volumes that use concatenated or striped plexes and contain data other than a file system. Use `raid5` for volumes that use a RAID5 plex regardless of the contents of the volume. Default: `fsgen`.

Table 2-3 describes the benefits and trade-offs of various plex layouts and lists scenarios where one plex type might provide better performance, or be more cost effective, than a different plex type. For optimal performance you might need to tune your system to the work load. The layout you choose depends on your specific system configuration, data availability and reliability needs, and application requirements.

Table 2-3: Choosing a Plex Type

Plex Type	Benefits and Possible Uses	Trade-offs
Concatenated	Allows you to use space on multiple disks that might otherwise be wasted. Concatenated plex can be mirrored for data redundancy. Good for volumes containing infrequently used data or data that does not change often or small volumes that can be confined to a single disk.	Possible uneven performance (hot spots, where one disk is in use by multiple applications). When mirrored, requires at least twice as much disk space (up to 32 times, depending on number of plexes).
Striped	Allows you to distribute data and therefore I/O load evenly across many disks. Good for: Large volumes that cannot be confined to a single disk. Applications with large read-request size. Volumes that contain data that changes often (many writes). Striping is preferred over RAID 5 in this case, because RAID 5 imposes the overhead of calculating and writing parity data along with volume data. Striped plexes can be mirrored for data redundancy and high availability.	When mirrored, requires at least twice as much disk space (up to 32 times, depending on number of plexes).
RAID 5	Provides redundancy through parity, using fewer disks than a volume with striped mirror plexes. Provides the I/O distribution benefit of striping. Good for volumes with a high read-to-write ratio.	Depending on the I/O stripe size, performance might be slower than a volume with striped plexes due to parity calculation. The RAID 5 plex type is not supported in a cluster.

The following sections provide worksheets to assist you in planning LSM volumes depending on the type of plex you want to use. Using the information in these worksheets will help you when you create volumes as described in Chapter 4.

Note

When you create an LSM volume with the volassist command (the recommended and simplest method), LSM performs all the necessary calculations and creates a volume and log plexes of the appropriate sizes. The following worksheets are provided to help you approximate the space needed and ensure the disk group has enough space for the volumes you want.

2.1.1 Planning an LSM Volume That Uses a Concatenated Plex

Use the following worksheet to plan an LSM volume that uses a concatenated plex.

Figure 2-1: Worksheet for Planning a Volume with Concatenated Plexes

Attribute	Default Values	Chosen Values
Volume name	No default
Volume size	No default
Number of data plexes	1
If more than one plex, DRL plex size	65 blocks for volumes less than or equal to 1 GB ^{[Footnote 1]}
Disk group name	rootdg
Usage type	`fsgen`
Total space required	(Volume size * number of plexes) + DRL size

2.1.2 Planning an LSM Volume That Uses a Striped Plex

Use the following worksheet to plan an LSM volume that uses a striped plex.

Figure 2-2: Worksheet for Planning a Volume with Striped Plexes

Attribute	Default Values	Chosen Values
Volume name	No default
Volume size	No default
Data unit size	64 KB
Number of columns	Minimum of two, based on number of disks in disk group and the volume size
Number of data plexes	1
If more than one plex, DRL plex size	65 blocks for volumes less than or equal to 1 GB ^{[Footnote 2]}
Disk group name	rootdg
Usage type	`fsgen`
Total space required	(Volume size * number of plexes) + DRL size

2.1.3 Planning an LSM Volume That Uses a RAID 5 Plex

Use the following worksheet to plan an LSM volume that uses a RAID 5 plex.

Figure 2-3: Worksheet for Planning a Volume with a RAID 5 Plex

Attribute	Default Values	Chosen Values
Volume name	No default
Volume size	No default
Data unit size	16 KB
Number of columns (NCOL)	Between 3 and 8 based on number if disks in disk group and the volume size	(Minimum of three)
Log plex size	10 * (data unit size * number of columns)
Disk group name	rootdg
Usage type	Must be `raid5`	`raid5`
Total space required	(Volume size * NCOL / (NCOL-1)) + log plex size

2.2 Planning Disk Groups

At a minimum, you must plan the rootdg disk group, which is created when you install LSM. Planning a disk group requires that you identify:

The space requirements for the disk group by identifying the size of volumes that you will create in the disk group, as described in Section 2.1.

Unused storage devices to meet the space requirement of the disk group, as described in Section 2.3.

When you plan a disk group, consider the following:

You must identify at least one unused storage device for the rootdg disk group when you install LSM.

A disk group should have more than one storage device to ensure that there are multiple copies of the disk group's configuration database.

To improve performance, keep the number of disks in rootdg to ten or fewer. Create additional disk groups if you:
- Have more than ten disks to place under LSM control.
- Might move the disk group to a different system.

A disk group should have storage devices on different buses to improve performance and availability.

Choose a name for the disk group carefully. There is no direct method for renaming disk groups, as there is for renaming volumes or LSM disks. To rename a disk group, you must deport it and import it with a new name. This requires stopping all activity on volumes in that disk group for the time necessary to deport and import it, which might not be acceptable or feasible in your environment.

The following considerations apply specifically to clusters:
- To improve performance, use the rootdg disk group only for system-related volumes such as the cluster_root domain and the swap devices for members. Try to keep the number of disks in rootdg to ten or fewer.
- The disks in each disk group should be accessible by all cluster members (on a shared bus) so that all members have access to the volumes, even if one or more members are not running. If some disks are accessible only to some members, try to use those disks for data that is needed only by the member directly attached; for example, that member's swap space.

Use the worksheets in Figure 2-4 and Figure 2-5 to plan disk groups. You can make copies and fill in the information on the copies rather than in the manual. This lets you keep the disk group information with each system running LSM, for your reference. Also, because you can change your LSM configuration at any time, you can make a new copy of the blank worksheets to record your changes.

In the appropriate worksheet, enter the following:

Under Disk Group Information, include any information that will help you keep track of the purpose of that disk group. For example, you might create a disk group called finance whose purpose is to contain one or more volumes that will be used by a financial application. You might create another disk group called db1, which will contain a volume used by a database.

Under Volume, Plex and Spare Disk Information, include the names of all volumes in that disk group, their plex type, which disks belong to which plex and identify any spare disks that will be used to replace failed disks. See Section 3.4.4 and Section 3.4.4.1 for more information on spare disks.
Figure 2-6 is an example of a completed worksheet.

Figure 2-4: Worksheet for Planning the rootdg Disk Group

Disk Group Information	Disks in Group	Bus/LUN Number	Disk Size	Volume, Plex, and Spare Disk Information
Name: rootdg Purpose:

Figure 2-5: Worksheet for Planning Additional Disk Groups

Disk Group Information	Disks in Group	Bus/LUN Number	Disk Size	Volume, Plex, and Spare Disk Information
Name: Purpose:

Figure 2-6 shows a consolidated example of what your disk group planning worksheets might look like when complete. Note that this example applies only to a standalone system, not a cluster.

Figure 2-6: Worksheet for Planning Disk Groups for a Standalone System (Consolidated Example)

Disk Group Information	Disks in Group	Bus/LUN Number	Disk Size	Volume, Plex, and Spare Disk Information
Name: rootdg Purpose: root file system and system disks.	dsk0	0	4 GB	root disk (encapsulated: rootvol plex-01)
	dsk1	0	4 GB	rootvol plex-02
	dsk4	2	4 GB	swapvol plex-01
	dsk5	2	4 GB	swapvol plex-02
	dsk16	6	4 GB	hot-spare disk
Name: data_dg Purpose: Database, must be redundant. Contains volume with mirrored striped plexes and DRL.	dsk6	3	18 GB	volume: db_vol plex: db_vol-01
	dsk7	3	18 GB	plex: db_vol-01
	dsk8	4	18 GB	plex: db_vol-02
	dsk9	4	18 GB	plex: db_vol-02
	dsk10	5	18 GB	plex: db_vol-03 (DRL plex)
	dsk11	5	18 GB	hot-spare disk
	dsk15	6	18 GB	hot-spare disk
Name: finance_dg Purpose: Financial application, must be highly available. Contains volume with RAID 5 plex (read-only application).	dsk20	7	9 GB	volume: fin_vol column: 1
	dsk25	8	9 GB	column 2
	dsk30	9	9 GB	column 3
	dsk35	10	9 GB	column 4
	dsk40	11	9 GB	column 5
	dsk45	16	9 GB	log plex
	dsk16	6	18 GB	hot-spare disk

2.3 Identifying Unused Storage Devices

Unused storage devices are unused disks, partitions, and RAID disks that LSM can initialize to become LSM disks for use in the rootdg disk group or in the other disk groups that you create.

You can also identify unused LSM disks for use in a disk group. An unused LSM disk is a storage device that you initialized for use by LSM but did not assign to a disk group.

The following sections describe how to identify unused disks, partitions, and LSM disks. See your hardware RAID documentation for information on identifying unused hardware RAID disks.

To identify unused storage devices, you can use:

The Disk Configuration Graphical User Interface (GUI) (Section 2.3.1).

The command-line interpreter interface (Section 2.3.2).

The voldisk list command on a system where LSM is running (Section 2.3.3).

2.3.1 Using the Disk Configuration GUI to Identify Unused Disks

To identify unused disks using the Disk Configuration GUI, start the Disk Configuration interface using either of the following methods:

From the system prompt, enter:
```
# /usr/sbin/diskconfig
```

From the SysMan Applications pop-up menu on the CDE Front Panel:
1. Choose Configuration.
2. Double click the Disk icon in the SysMan Configuration folder.
A window titled Disk Configuration on hostname is displayed. This is the main window for the Disk Configuration GUI, and lists the following information for each disk:
- The disk name, such as dsk10
- The device model, such as RZ1CB-CA
- The bus number for the device
For more information about a disk, double click on the list item (or click Configure when a disk is highlighted). The Disk Configuration: Configure Partitions: window is displayed.
This window contains:
- A graphical representation of the disk partitions in a horizontal bar-chart format and disk information such as the disk name, the total size of the disk, and usage information.
- A Partition Table button that you can click to display a bar chart of the current partitions in use, their sizes, and the file system in use.
- A Disk Attributes button that you can click to display values for disk attributes.

For more information about the Disk Configuration GUI, see its online help.

2.3.2 Using Operating System Commands to Identify Unused Disks

You can use the following operating system commands to identify unused disks:

List all the disks on the system:

# file /dev/rdisk/dsk*c

Information similar to the following is displayed:

/dev/rdisk/dsk0c:       character special (19/38) SCSI #1 "RZ1CD-CS" disk #1 (SCSI ID #0) (SCSI LUN #0)
/dev/rdisk/dsk10c:      character special (19/198) SCSI #3 "RZ1CD-CS" disk #3 (SCSI ID #5) (SCSI LUN #0)
/dev/rdisk/dsk11c:      character special (19/214) SCSI #4 "RZ1BB-CS" disk #4 (SCSI ID #0) (SCSI LUN #0)
/dev/rdisk/dsk12c:      character special (19/230) SCSI #4 "RZ1BB-CS" disk #5 (SCSI ID #1) (SCSI LUN #0)
/dev/rdisk/dsk13c:      character special (19/246) SCSI #4 "RZ1BB-CS" disk #6 (SCSI ID #2) (SCSI LUN #0)
/dev/rdisk/dsk14c:      character special (19/262) SCSI #4 "RZ1BB-CS" disk #7 (SCSI ID #3) (SCSI LUN #0)
/dev/rdisk/dsk15c:      character special (19/278) SCSI #4 "RZ1CD-CS" disk #8 (SCSI ID #4) (SCSI LUN #0)
/dev/rdisk/dsk16c:      character special (19/294) SCSI #4 "BD009635C3" disk #9 (SCSI ID #5) (SCSI LUN #0)
/dev/rdisk/dsk17c:      character special (19/310) SCSI #4 "BD009635C3" disk #10 (SCSI ID #6) (SCSI LUN #0)
/dev/rdisk/dsk18c:      character special (19/326) SCSI #5 "RZ1CD-CS" disk #11 (SCSI ID #0) (SCSI LUN #0)
/dev/rdisk/dsk19c:      character special (19/342) SCSI #5 "RZ1BB-CS" disk #12 (SCSI ID #1) (SCSI LUN #0)
/dev/rdisk/dsk1c:       character special (19/54) SCSI #1 "RZ1BB-CA" disk #2 (SCSI ID #2) (SCSI LUN #0)
/dev/rdisk/dsk20c:      character special (19/358) SCSI #5 "RZ1CB-CA" disk #13 (SCSI ID #2) (SCSI LUN #0)
/dev/rdisk/dsk21c:      character special (19/374) SCSI #5 "RZ1CB-CA" disk #14 (SCSI ID #3) (SCSI LUN #0)
/dev/rdisk/dsk22c:      character special (19/390) SCSI #5 "RZ1CF-CF" disk #15 (SCSI ID #4) (SCSI LUN #0)
/dev/rdisk/dsk23c:      character special (19/406) SCSI #5 "RZ1CF-CF" disk #8 (SCSI ID #5) (SCSI LUN #0)
/dev/rdisk/dsk24c:      character special (19/422) SCSI #5 "BD009635C3" disk #9 (SCSI ID #6) (SCSI LUN #0)
/dev/rdisk/dsk25c:      character special (19/438) SCSI #6 "RZ1BB-CS" disk #10 (SCSI ID #1) (SCSI LUN #0)
/dev/rdisk/dsk26c:      character special (19/454) SCSI #6 "RZ1CD-CS" disk #11 (SCSI ID #3) (SCSI LUN #0)
/dev/rdisk/dsk27c:      character special (19/470) SCSI #6 "RZ1CD-CS" disk #12 (SCSI ID #5) (SCSI LUN #0)
/dev/rdisk/dsk2c:       character special (19/70) SCSI #1 "RZ1CD-CS" disk #3 (SCSI ID #4) (SCSI LUN #0)
/dev/rdisk/dsk3c:       character special (19/86) SCSI #1 "RZ1CD-CS" disk #4 (SCSI ID #6) (SCSI LUN #0)
/dev/rdisk/dsk4c:       character special (19/102) SCSI #2 "RZ1BB-CS" disk #5 (SCSI ID #0) (SCSI LUN #0)
/dev/rdisk/dsk5c:       character special (19/118) SCSI #2 "RZ1CD-CS" disk #6 (SCSI ID #2) (SCSI LUN #0)
/dev/rdisk/dsk6c:       character special (19/134) SCSI #2 "RZ1CD-CS" disk #7 (SCSI ID #4) (SCSI LUN #0)
/dev/rdisk/dsk7c:       character special (19/150) SCSI #2 "RZ1CD-CS" disk #0 (SCSI ID #6) (SCSI LUN #0)
/dev/rdisk/dsk8c:       character special (19/166) SCSI #3 "RZ1BB-CA" disk #1 (SCSI ID #1) (SCSI LUN #0)
/dev/rdisk/dsk9c:       character special (19/182) SCSI #3 "RZ1CD-CS" disk #2 (SCSI ID #3) (SCSI LUN #0)

To verify if a disk or partition is unused, choose a disk from the output of the file /dev/rdisk/dsk*c command and enter the disklabel command with the name of the disk; for example:

# disklabel dsk20c

Disk partition information similar to the following is displayed:

type: SCSI
disk: RZ1CB-CA
label:
flags: dynamic_geometry
bytes/sector: 512
sectors/track: 113
tracks/cylinder: 20
sectors/cylinder: 2260
cylinders: 3708
sectors/unit: 8380080
rpm: 7200
interleave: 1
trackskew: 9
cylinderskew: 9
headswitch: 0           # milliseconds
track-to-track seek: 0  # milliseconds
drivedata: 0
 
8 partitions:
#            size       offset    fstype  fsize  bsize   cpg  # ~Cyl values
  a:       131072            0    unused      0      0        #      0 - 57*
  b:       262144       131072    unused      0      0        #     57*- 173*
  c:      8380080            0    unused      0      0        #      0 - 3707
  d:            0            0    unused      0      0        #      0 - 0
  e:            0            0    unused      0      0        #      0 - 0
  f:            0            0    unused      0      0        #      0 - 0
  g:      3993432       393216    unused      0      0        #    173*- 1940*
  h:      3993432      4386648    unused      0      0        #   1940*- 3707

See the disklabel(8) reference page for more information on the disklabel command.

If you are using AdvFS, display the disks in use by all domains:

# ls /etc/fdmns/*/*
/etc/fdmns/cluster_root/dsk7b   /etc/fdmns/root2_domain/dsk11a
/etc/fdmns/cluster_usr/dsk7g    /etc/fdmns/root_domain/dsk1a
/etc/fdmns/cluster_var/dsk7h    /etc/fdmns/usr_domain/dsk1g
/etc/fdmns/root1_domain/dsk10a

If you are using UFS, display all mounted file sets:
```
# mount
```

2.3.3 Using the LSM voldisk Command to Identify Unused Disks

When LSM starts, it obtains a list of disk device addresses from the operating system software and checks the disk labels to determine which devices are initialized for LSM use and which are not.

If LSM is running on the system, you can use the voldisk command to display a list of all known disks and to display detail information about a particular disk:

To view a list of disks, enter:

# voldisk list

Information similar to the following is displayed.

DEVICE       TYPE      DISK         GROUP        STATUS  
dsk0         sliced    -            -            unknown 
dsk1         sliced    -            -            unknown 
dsk2         sliced    dsk2         rootdg       online  
dsk3         sliced    dsk3         rootdg       online  
dsk4         sliced    dsk4         rootdg       online  
dsk5         sliced    dsk5         rootdg       online  
dsk6         sliced    dsk6         dg1          online  
dsk7         sliced    -            -            online  
dsk8         sliced    dsk8         dg1          online  
dsk9         sliced    -            -            online  
dsk10        sliced    -            -            online  
dsk11        sliced    -            -            online  
dsk12        sliced    -            -            online  
dsk13        sliced    -            -            unknown 
dsk14        sliced    -            -            unknown

The following list describes the information in the output:

`DEVICE`	Specifies the disk access name assigned by the operating system.
`TYPE`	Specifies the LSM disk type (sliced, simple, or nopriv).
`DISK`	Specifies the LSM disk media name. A dash (-) means the device is not assigned to a disk group and therefore does not have an LSM disk media name.
`GROUP`	Specifies the disk group to which the device belongs. A dash (-) means the device is not assigned to a disk group.
`STATUS`	An unused storage device is one that does not have a DISK name or GROUP name and has a status of unknown. An unused LSM disk is one that has a DISK name but has no GROUP name and a status of online or offline.

To display detail information about an LSM disk, enter:

# voldisk list disk

The following example displays information for an LSM disk called dsk5:

Device:    dsk5
devicetag: dsk5
type:      sliced
hostid:    servername
disk:      name=dsk5 id=942260116.1188.servername
group:     name=dg1 id=951155418.1233.servername
flags:     online ready autoimport imported
pubpaths:  block=/dev/disk/dsk5g char=/dev/rdisk/dsk5g
privpaths: block=/dev/disk/dsk5h char=/dev/rdisk/dsk5h
version:   n.n
iosize:    min=512 (bytes) max=2048 (blocks)
public:    slice=6 offset=16 len=2046748
private:   slice=7 offset=0 len=4096
update:    time=952956192 seqno=0.11
headers:   0 248
configs:   count=1 len=2993
logs:      count=1 len=453
Defined regions:
 config   priv     17-   247[   231]: copy=01 offset=000000 enabled
 config   priv    249-  3010[  2762]: copy=01 offset=000231 enabled
 log      priv   3011-  3463[   453]: copy=01 offset=000000 enabled

The size of an LSM disk is displayed in blocks as the len= value in the public: row. 2048 blocks equal 1 MB.

See the voldisk(8) reference page for more information on the voldisk command.