Physical Storage Management
Module 6 Data ONTAP 7.3 Fundamentals
Module Objectives
By the end of this module, you should be able to: Describe Data ONTAP RAID technology Identify a disk in a disk shelf based on its ID Execute commands to determine disk ID Identify a hot-spare disk in an FAS system Calculate usable disk space Describe the effects of using multiple disk types Execute aggregate commands in Data ONTAP Define and create an aggregate
2008 NetApp. All rights reserved. 2
Disks
2008 NetApp. All rights reserved.
Disks
All data is stored on disks To understand how physical media is managed in your storage system, we will address:
Disk types Disk qualification Disk ownership Spare disks
2008 NetApp. All rights reserved.
Supported Disk Topologies
FC-AL
FAS2000
FAS3000
FAS6000
SATA
FAS2000
FAS3000
FAS6000
R200
2008 NetApp. All rights reserved.
Disk Qualification
Use only NetApp Qualified Disks
Modifying the Disk Qualification Requirement file can cause your storage system to halt.
2008 NetApp. All rights reserved.
Disk Ownership
Disks must be assigned to (owned by) a controller.
Software Disk Ownership Ownership is assigned FAS270 FAS3000 series FAS2000 series FAS6000 series Hardware Disk Ownership Ownership is based on slot used R200 FAS250 FAS3000 series
2008 NetApp. All rights reserved.
Disk Ownership
system> sysconfig -r Volume vol0 (online, normal) (block checksums) Plex /vol0/plex0 (online, normal, active) RAID group /vol0/plex0/rg0 (normal) RAID Disk parity data Device 3a.16 3b.17 HA 4a 4a SHELF 1 1 BAY 0 1 CHAN FC:A FC:A Used (M... 17000/... 17000/...
Disk ID = Loop_id.Device_id
2008 NetApp. All rights reserved.
Disk Ownership: Loop_id
The loop_id is the designation for the slot and port where an adapter is located. In the following illustration, the loop_id is 3b.
3b
PROPERLY SHUT DOWN SYSTEM BEFORE OPENING CHASSIS. DISCONNECT AC POWER CORD BEFORE REMOVAL PCI 1 PCI 3 DISCONNECT AC POWER CORD BEFORE REMOVAL
I O PCI 2 PCI 4
I O
Console
L I N K
e0a 0a
F C
e0b
RLM
e0c
e0d
L I N K
0b
L I N K
0c
0d
L I N K
0e
LVD SCSI AC AC
L I
L I
L I
Console
N K
0a
0b
N K
e0a
e0b
RLM
e0c
e0d
N K
0c
0d
N K
0e
LVD SCSI
status
status
2008 NetApp. All rights reserved.
Disk Ownership: Device_id
Shelf ID
13 12 11 10 9 8 7 6 5 4 3 2 1 0 Bay Number Device ID 2916 4532
Shelf ID 1 2
Bay Number 130 130
3
4 5 6
130
130 130 130
6148
7764 9380 10996
7
2008 NetApp. All rights reserved.
130
125112
10
The fcstat device_map Command
Use the fcstat command to troubleshoot disks and shelves. Use the fcstat device_map command to show disks and their relative physical position map of drives on an FC loop.
2008 NetApp. All rights reserved.
11
Matching Disk Speeds
When creating an aggregate or traditional volume, Data ONTAP selects disks:
With same speed That match speed of existing disks
Data ONTAP verifies that adequate spares are available
2008 NetApp. All rights reserved.
12
Using Multiple Disk Types in an Aggregate
Drives in an aggregate can be:
Different speeds On the same shelf or on different shelves
Avoid mixing drive types within an aggregate The spares pool is global
12
34
56
2008 NetApp. All rights reserved.
13
Spare Disks
What is the purpose of spare disks?
Increase aggregate capacity Replace failed disks Zeros disk automatically when the disk is brought into use
It is best to zero drives in the spares pool in advance, allowing Data ONTAP to use the drives immediately
system> disk zero spares
2008 NetApp. All rights reserved.
14
Sizing
2008 NetApp. All rights reserved.
15
Disk Sizing
To properly provision NetApp storage systems, you must know how disk sizes are calculated: All disks are right-sized Count the size of data disks, not parity disks
NOTE: The df command does not reflect parity disks.
Use df -h to view the output in a format you can read
2008 NetApp. All rights reserved.
16
Right-Sizing
Disk Type FC/SCSI Disk Size 72 GB Right-Sized Capacity 68 GB Available Blocks 139,264,000
144 GB
300 GB ATA/SATA 160 GB 250 GB 320 GB 500 GB
136 GB
272 GB 136 GB 212 GB 274 GB 423 GB
278,528,000
557,056,000 278,258,000 434,176,000 561,971,200 866,531,584
NOTE: ATA drives have only 512 bytes per sector and lose an additional 1/9 or 12.5% due to block checksum allocation.
2008 NetApp. All rights reserved.
17
Usable Disk Space
When disks are right-sized, 10% of the space is reserved for WAFL. This reserved space is:
Used for system and core usage to maximize disk efficiency Similar to other operating systems (for example, the UNIX FFS)
The space that remains after right-sizing is usable disk space, which can be used for either:
Traditional volumes Aggregates with flexible volumes
2008 NetApp. All rights reserved. 18
Disk Space Allocation: Aggregates
Aggregates with a Traditional VolumeEach aggregate has 10% allocated for WAFL. Traditional VolumesEach volume has 20% allocated for Snapshot reserve. The remainder is used for client data. Snapshot ReserveThe amount of space allocated for Snapshot reserve is adjustable. To use this space for data (not recommended), you must manually override the allocation used for Snapshot copies.
2008 NetApp. All rights reserved.
Aggregate Space
10%
WAFL Overhead
WAFL Aggregate Space
80%
90%
Snapshot Reserve
20%
(Adjustable)
19
Disk Space Allocation: Flexible Volumes
AggregatesEach aggregate has 5% allocated for Snapshot reserve and 10% allocated for WAFL. Flexible VolumesEach volume has 20% allocated for Snapshot reserve. The remainder is used for client data. Snapshot ReserveThe amount of space allocated for Snapshot reserve is adjustable. To use this space for data (not recommended), you must manually override the allocation used for Snapshot copies.
2008 NetApp. All rights reserved.
Aggregate Space
10%
WAFL Overhead WAFL Aggregate Space
FlexVol Space Plus Aggregate Snapshot Reserve
FlexVol1x
80%
90%
.snapshot
20%
95%
FlexVol#n
80%
.snapshot
20%
Aggregate Snapshot Reserve
5%
(Adjustable)
20
Disk Protection
2008 NetApp. All rights reserved.
21
Disk Protection
Data ONTAP protects disks through:
RAID Disk scrubbing
Data ONTAP can assist in recovering from disk failures
2008 NetApp. All rights reserved.
22
RAID Groups
RAID groups are a collection of data disks and parity disks RAID groups provide protection through parity Data ONTAP organizes disks into RAID groups Data ONTAP supports:
RAID 4 RAID-DP
2008 NetApp. All rights reserved.
23
RAID 4 Technology
RAID 4 protects against data loss that results from a single-disk failure in a RAID group A RAID 4 group requires a minimum of two disks:
One parity disk One data disk
Parity
Data
Data
Data
Data
Data
Data
Data
2008 NetApp. All rights reserved.
24
RAID-DP Technology
RAID-DP protects against data loss that results from double-disk failures in a RAID group A RAID-DP group requires a minimum of three disks:
One parity disk One double-parity disk One data disk
Parity
DoubleParity
Data
Data
Data
Data
Data
Data
2008 NetApp. All rights reserved.
25
RAID Group Size
RAID-DP NetApp Platform All storage systems (with SATA disks) All storage systems (with FC disks) Minimum Group Size 3 3 Maximum Group Size 16 28 Default Group Size 14 16
RAID 4 NetApp Platform FAS270 All other storage systems (with SATA) All other storage systems (with FC) Minimum Group Size 2 2 2 Maximum Group Size 14 7 14 Default Group Size 7 7 8
2008 NetApp. All rights reserved.
26
Data Reliability
RAID-level checksums enhance data protection and reliability Two processes:
options raid.media_scrub
Checks for media errors only If enabled, runs continuously in the background
options [Link] (also called disk scrubbing)
Checks for media errors Corrects parity consistency
2008 NetApp. All rights reserved.
27
RAID Checksums
Zone Checksums (ZCS)
Eight 512-byte sectors (4,096 bytes) per block Every 64th block checksums the previous 63 WAFL never uses these checksum blocks; RAID does Available for V-Series Eight 512-byte sectors (4096 bytes) per block Every sector checksums itself Are faster than ZCS Are the standard for FC, SCSI and V-Series disks 8/9ths ATA disks; every ninth sector checksums the previous eight
Block Checksums (BCS)
2008 NetApp. All rights reserved.
28
Comparing Media and RAID Scrubs
A media scrub: Is always running in the background when the storage system is not busy Looks for unreadable blocks at the lowest level (0s and 1s) Is unaware of the data stored in a block Takes corrective action when it finds too many unreadable blocks on a disk (sends warnings or fails a disk, depending on findings) A RAID scrub: Is enabled by default Can be scheduled or disabled Disabling is not recommended Uses RAID checksums Reads a block and then checks the data If the RAID scrub finds a discrepancy between the RAID checksum and the data read, it re-creates the data from parity and writes it back to the block Ensures that data has not become stale by reading every block in an aggregate, even when users havent accessed the data
29
2008 NetApp. All rights reserved.
About Disk Scrubbing
Automatic RAID scrub:
By default, begins at 1 a.m. on Sundays Schedule can be changed by an administrator Duration can be specified by an administrator
Manual RAID scrub overrides automatic settings
To scrub disks manually:
[Link] off And then: aggr scrub start
To view scrub status:
aggr scrub status aggr_name
2008 NetApp. All rights reserved. 30
RAID Group Options
options [Link] options [Link].perf_impact options [Link] options [Link].perf_impact vol options <volname> raidtype aggr options <aggrname> raidtype
NOTE: For a complete list of RAID options, see your product documentation.
2008 NetApp. All rights reserved.
31
disk Commands
disk disk disk disk disk disk disk disk fail diskname remove diskname swap unswap replace [start|stop] zero spares scrub [start|stop] sanitize
2008 NetApp. All rights reserved.
32
Disk Failures
Volume 1
Raid Group 0
Spares
Parity Data Data Data
Spares are global.
2008 NetApp. All rights reserved.
33
Degraded Mode
Degraded mode occurs when: A single disk fails in a RAID 4 group with no spares Two disks fail in a RAID-DP group with no spares Degraded modes operates for 24 hours, during which time: Data is still available Performance is less-than-optimal
Data must be recalculated from the parity until the failed disk is replaced CPU usage increases to calculate from parity
System shuts down after 24 hours To change time interval, use the options [Link] command If an additional disk in the RAID group fails during degraded mode, the result will be data loss
2008 NetApp. All rights reserved. 34
Replacing a Failed Disk by Hot Swapping
Hot-swapping is the process of removing or installing a disk drive while system is running and allows for:
Minimal interruption The addition of new disks as needed
Removing two disks from a RAID 4 group:
Double-disk failure Data loss will occur
Removing two disks from a RAID-DP group:
Degraded mode No data loss
2008 NetApp. All rights reserved. 35
Replacing Failed Disks
750 GB
1 TB
750 GB
750 GB
750 GB
750 GB
NOTE: Disk resizing occurs if a smaller disk is replaced by a larger one.
2008 NetApp. All rights reserved.
36
Aggregates
2008 NetApp. All rights reserved.
37
Aggregates
Aggregates represent physical storage: Made up of one or more RAID groups A RAID group includes:
One or more data disks Parity disks
RAID 4 has only one parity disk RAID-DP has two parity disks Data is striped for parity protection
A flexible volume depends on an aggregate for physical storage
2008 NetApp. All rights reserved. 38
Naming Rules for Aggregates
An aggregate name must:
Begin with either a letter or the underscore character (_) Contain only letters, digits, and underscore characters (_) Contain no more than 255 characters
2008 NetApp. All rights reserved.
39
Adding an Aggregate
To add an aggregate using the CLI:
aggr create
To add an aggregate using FilerView, use the Aggregate Wizard When adding aggregates, you must have the following information available:
Aggregate name Parity (DP is default) RAID group size (minimum) Disk selection method Disk size Number of disks (including parity)
40
2008 NetApp. All rights reserved.
Creating an Aggregate Using the CLI
The following is an example of a CLI entry used to create an aggregate:
system> aggr create aggr_name 24
Creates an aggregate called aggr_name with 24 disks By default, this aggregate uses RAID-DP Using the default, 4 of the 24 disks are parity drives
2008 NetApp. All rights reserved.
41
Common Aggregate Commands
aggr create <aggrname> [options] <disklist> aggr add <aggrname> [options] <disklist> aggr status <aggrname> [options] aggr rename <aggrname> <new-aggrname> aggr show_space [-b] <aggrname> aggr offline {<aggrname> | <plexname>} aggr online {<aggrname> | <plexname>} aggr destroy {<aggrname> | <plexname>}
2008 NetApp. All rights reserved.
42
Creating an Aggregate Using the FilerView Aggregate Wizard
2008 NetApp. All rights reserved.
43
Aggregate Size
In Data ONTAP prior to version 7.3: Aggregate size is calculated using:
sysconfig r
All disks in the aggregate (parity and data) are included In Data ONTAP 7.3: Aggregate size is calculated using the size of data disks Only data disks in the aggregate are included (parity disks are excluded)
2008 NetApp. All rights reserved. 44
Module Summary
In this module, you should have learned to: NetApp supports FC-AL and SATA disk drives Use either FilerView or the CLI to find disk information Data ONTAP organizes disks into RAID groups RAID groups consist of data disks and parity disks Degraded mode occurs when a single disk fails in a RAID 4 group with no spares, or two disks fail in a RAID-DP group with no spares
2008 NetApp. All rights reserved. 45
Exercise
Module 6: Physical Storage Management Estimated Time: 60 minutes
Answers
Module 6: Physical Storage Management
Check Your Understanding
What is a RAID group?
A collection of disks organized to protect data that includes:
One or more data disks Data striped for performance One or two parity disks for protection
Why use double parity?
To protect against a double-disk failure
2008 NetApp. All rights reserved.
48
Check Your Understanding (Cont.)
How many RAID groups does the following command create? aggr create newaggr 32
Assuming a default RAID group size of 16, this creates two RAID groups
What is the minimum size of a RAID-DP group?
Three disks (one data, two parity)
2008 NetApp. All rights reserved.
49
