RAIDCore User Manual P/N Revision A December 2009

(1)

(2)

(3)

Figures

1 Dot Hill® window, showing Install Windows Management Suite selected . . . 40

2 License Agreement window, showing I accept the terms of the License Agreement selected. . . 40

3 Four-Drive RAID 5 Array at the Main Menu (typical window). . . 57

4 Example of a RAID 5 Swapped for a RAID 10 (typical window) . . . 59

5 Controller Options Section of the RAIDCore(tm) Array Configuration Window (typical) . . . 61

6 First Time Run Dialog Box (typical) . . . 64

7 RAIDConsole Dot Hill Window (typical) . . . 66

8 Array Status Window: Accessing Array Properties . . . 67

9 RAIDConsole Properties Window (typical). . . 67

10 Sections of the Array Status Window (typical) . . . 68

11 Icons of the Array View Section of the Array Status Window (typical) . . . 69

12 Array View, Array Status Window: Select an Array to be Copied . . . 80

13 Array View: Un-Link Command . . . 83

14 RAIDConsole - Event Service Window: Rescan Button. . . 84

15 RAIDConsole Window: Task -> Pause and Task -> Remove Options. . . 87

16 License Window, showing the license level and the space for entering an Activation ID . . . 94

17 An example of an Entitlement Certificate, which contains the Activation ID . . . 95

18 Select Display Columns Window: Edit Columns . . . 97

19 Notification Setup Window: Notification Events . . . 100

20 Options Drop-Down Menu of the SMART Information Window . . . 102

(10)

(11)

Tables

1 Document Conventions . . . 11

2 System Requirements for RAIDCore . . . 15

3 Features of RAIDCore . . . 16

4 Feature Set for RAIDCore: by RAIDCore License Level . . . 21

5 RAID Levels and Selection Criteria . . . 23

6 RAID Levels - General Characteristics. . . 23

7 Array States . . . 25

8 Failure States by RAID Level . . . 25

9 Array Expansion Considerations . . . 26

10 Types of Tasks per Array . . . 27

11 Disk States . . . 28

12 Minimum Disk Configurations per RAID Level . . . 28

13 Sparing Options . . . 29

14 System Setup Process . . . 33

15 Driver Files for Windows Operating Systems. . . 34

16 Windows Operating System: Install Drivers, Configuration Devices, and Management Suite. . . 37

17 Components of the Management Suite . . . 39

18 Installing the Red Hat Linux Drivers . . . 42

19 SUSE Linux 10.3: Installing the SUSE Linux Drivers . . . 45

20 SUSE Linux 11: Installing the SUSE Linux Drivers . . . 47

21 Linux Operating System: Install Drivers, Configuration Devices, and Management Suite. . . 49

22 Components Installed with the Management Suite (Linux) . . . 51

23 Reasons for Using the RAIDCore BIOS Configuration Utility (Option ROM) . . . 53

24 BIOS Color Codes . . . 53

25 Selecting An Array Size . . . 56

26 Choices for Selecting a Boot Array . . . 62

27 Elements at the Array View Section of the Array Status Window . . . 69

28 Disk List Section of the Array Status Window . . . 70

29 Array List Section of the Array Status Window. . . 71

30 Event View Elements, Array Status Window . . . 72

31 I/O Status Bar Elements, Array Status Window . . . 72

32 Restrictions and Conditions When Transforming a Legacy Disk . . . 75

33 Issues and Recommendations for Creating and Formatting Arrays. . . 76

34 Issues and Recommendations for Transforming Arrays . . . 78

35 Issues and Recommendations for Copying Arrays . . . 80

36 Options for Unlinking Copied Arrays . . . 82

37 Issues and Recommendations for Deleting Arrays . . . 85

38 Cache Array Options . . . 86

39 Consistency Check Options . . . 89

40 Splitting a Mirror by RAID Level . . . 90

41 Event Notification Types. . . 99

42 Event Log Priority Levels . . . 99

43 System Does Not Boot . . . 105

44 The BIOS Configuration Utility Does Not Display. . . 106

45 Warning Messages: POST Screen . . . 106

46 Troubleshooting: Cannot Create an Array . . . 109

47 Troubleshooting: An Array is in an Offline State . . . 109

48 Troubleshooting: An Array is in a Critical State . . . 110

49 Troubleshooting: Cannot Assign a Dedicated Hot Spare to an Array . . . 110

50 Troubleshooting: Cannot Create a Global Hot Spare. . . 111

51 Troubleshooting: Cannot Create an Array Larger Than 2 TB . . . 112

(12)

56 Troubleshooting: Cannot Initialize a Disk . . . 114

57 Modes for the bcadm Program. . . 119

58 bcadm Controller List Elements . . . 120

59 bcadm Disk List Elements . . . 121

60 bcadm Array List Elements . . . 122

(13)

Preface

This user guide:

• Provides information about arrays, disks, and RAID levels (RAID types).

• Describes how to improve storage system performance or reliability by understanding array and disk tasks and options.

Intended audience

This user guide is intended for use by system administrators and technicians who are experienced with the following:

• Direct Attached Storage (DAS), Storage Area Network (SAN), or Network Attached Storage (NAS) operators

• Network administration • Network installation

• Storage system installation and configuration

Prerequisites

Prerequisites for installing and configuring this product include familiarity with: • Servers and computer networks

• RAID and input/output signal technology (such as SCSI, SAS, or SATA) • Fibre Channel and Ethernet protocols

Document conventions and symbols

Table 1 Document Conventions

Convention Element

Navy blue text: Figure 1 Cross-reference links and e-mail addresses

Navy blue, underlined text (http://www.example.com)

Web site addresses

Bold font • Key names

• Text typed into a GUI element, such as into a box

• GUI elements that are clicked or selected, such as menu and list items, buttons, and check boxes

Italics font Text emphasis

Monospace font • File and directory names

• System output • Code

• Text typed at the command-line

Monospace, italic font • Code variables

• Command-line variables

Monospace, bold font Emphasis of file and directory names, system output, code, and

(14)

12 Preface

CAUTION: Indicates that failure to follow directions could result in damage to equipment or data.

IMPORTANT: Provides clarifying information or specific instructions.

NOTE: Provides additional information.

TIP: Provides helpful hints and shortcuts.

Customer Support

(15)

1 Chapter 1 Safety Precautions

General

This section includes general safety precautions and specific RAIDCore cautions. Read and keep this user manual for future reference.

Safety Definitions

CAUTION: A Caution denotes the possibility of minor to severe damage to equipment and/or loss of data.

Safety Precautions

CAUTION: Before installing RAIDCore read all instructions and cautionary markings on the equipment and all

appropriate sections of this user manual.

CAUTION: Before installing RAIDCore in an existing system, back up all critical data. Failure to follow this

accepted system management practice could result in loss of data.

CAUTION: All components (drivers and applications) must be installed when upgrading to the latest version

of RAIDCore. This means that an older version driver will not work with the latest version of the RAIDCore application and vice versa.

CAUTION: If a disk has a RAID array on it, it cannot be initialized. To initialize the disk anyway, delete the

array. Make sure this is what is desired. Deleted data on a disk cannot be retrieved.

CAUTION: Do not upgrade the system if there are any tasks (such as a Transform or a Restore) running on

any of the arrays.

CAUTION: Deleting an array permanently destroys all data that is on the array. This action cannot be

undone.

CAUTION: Booting an operating system from a RAIDCore bootable array is supported. Dot Hill Systems

Corp. does not support booting from a RAIDCore array without having a RAIDCore BIOS (option ROM).

CAUTION: Leaving Write Back caching enabled might increase the likelihood of data being corrupted,

if the system experiences a power interruption or unexpected shutdown.

CAUTION: When a task is cancelled, data that is in the part of the task that did not finish will be lost on the

(16)

(17)

2 Chapter 2 Getting Started

RAIDCore Technology

RAIDCore consists of (a) storage management and (b) RAID controller and port virtualization. RAIDCore runs on existing systems by using a motherboard’s built-in SATA ports or enhanced chipsets that have SAS disk or expansion capability.

RAIDCore provides a common interface to arrays, regardless of the type of controller.

Who Should Use This Manual

Only trained, experienced, and authorized personnel should install RAIDCore and use its features and capabilities.

All unit operators must be familiar with system hardware, data storage, RAID technology, input/output signal technology (such as SCSI, SAS, or SATA), and Direct Attached Storage (DAS) Storage, Network Attached Storage (NAS), and/or Storage Area Network (SAN) concepts and technology.

The intended user audience of this user manual are system administrators and experienced users.

System Requirements for Using RAIDCore

Make sure the systems that use RAIDCore meet the requirements indicated in the following table.

Supported Controllers

The following controllers are supported by the latest version of RAIDCore: • AMD SB0700, SP5100

Supported Operating Systems

The following operating systems are supported by the current release of RAIDCore:

• Microsoft Windows® Server 2003 SP2: Enterprise Edition, Standard Edition: 32-bit and 64-bit. • Microsoft Windows® Server 2008: Enterprise Edition, Standard Edition: 32-bit and 64-bit. • Microsoft Windows® Server 2008 R2: Enterprise Edition, Standard Edition: 64-bit. • Red Hat® Enterprise Linux® 5.3 and 5.4: 32-bit and 64-bit

• SUSE® Linux Enterprise Server 10.3 and 11: 32-bit and 64-bit

NOTE: Microsoft Windows® Server 2003 without SP2 is not supported by RAIDCore. Table 2 System Requirements for RAIDCore

Component Requirements

Memory (RAM) Minimum: 512 MB. Recommended: 2 GB.

(18)

16 Chapter 2 Getting Started

Features of RAIDCore

The features of RAIDCore described in this user manual apply to all license levels and supported operating systems. See Supported Operating Systems on page 15.

The following table describes these features. Also see Table 4 on page 21 for a summary of features that are available with each license level of RAIDCore: RAIDCore Basic, RAIDCore Plus, and RAIDCore VST.

Table 3 Features of RAIDCore

RAIDCore Feature Description

Arrays (general information) RAIDCore allows:

• Creating arrays of different RAID levels using the same disks.

NOTE: RAID 5 is supported by RAIDCore Plus and

RAIDCore VST. RAID 1n, 10n, and 50 are supported by RAIDCore VST.

• Creating different RAID level arrays on the same disk, to adapt each array to the I/O that it processes.

• Creating an array from a mix of different-type disks. For example, a RAID 10 array can be created from a group of disks that contain two SATA II HDDs, a SAS HDD, and a SATA SSD.

• Migrating an existing array to another RAID level, if the type of array being used is not the optimal type for the application. This function depends on the array capacity and redundancy level.

A array refers to data storage created by RAIDCore from one or more disks. Although an array can be created from several disks, it is seen by the operating system as a single disk.

Array copying The Copy option is used to create a new array that contains an exact copy of the data on a source array. It combines the Create Array and Copy Array functions into a single command.

NOTE: This feature is available only with RAIDCore

VST. See Copy Arrays (supported only by RAIDCore VST) on page 80.

Array hiding An array can be hidden from the operating system so that neither the software nor users can see or access it. This is a key element in creating secure data backups.

(19)

Background array initialization (BGI)

The background initialization of a redundant array creates the parity data that allows the array to maintain its redundant data and survive a disk failure.

Because background initialization helps RAIDCore identify and correct problems that might occur with the redundant data at a later time, background initialization is similar to check consistency. Background initialization allows a redundant array to be used immediately. Data is not lost if a disk goes offline prior to completion of the BGI process.

Cache support for arrays Various array-caching options are supported: No Cache, Read Cache, Write Back Cache, Read + Write Back Cache.

Cache support for disks Various disk-caching options are supported: No Cache, Disk Read Ahead, Disk Write Back, Disk Read Ahead + Write Back.

Consistency Check A Consistency Check is a background operation that verifies and corrects the mirror or parity data for fault-tolerant disks. It is recommended that a Consistency Check be run periodically on the disks.

By default, a Consistency Check automatically corrects mirror or parity inconsistencies.

A Consistency Check can be scheduled or started manually by the system user.

Controller spanning A unique feature of RAIDCore that allows very wide arrays to be created from disks connected to different controllers.

VST.

See Supported Controllers on page 15.

Disk roaming With disk roaming, SAS/SATA cables can be disconnected from their disks and shuffled without confusing RAIDCore.

NOTE: Disconnect the SAS/SATA cables from the disks

only when the system is offline. Disk roaming also allows:

• Disks to be moved to different slots in the backplane. RAIDCore detects which disks belong to which arrays, regardless of where the disks are moved in the backplane. • Disk(s) to be moved between systems.

Table 3 Features of RAIDCore (continued)

(20)

Fault tolerance The following fault tolerance features are available with RAIDCore, in order to prevent data loss in case of a failed disk.

• Disk failure detection (automatic).

• Array rebuild using hot spares (automatic, if the hot spare is configured for this functionality).

• Parity generation and checking (RAID 5 only).

• Hot-swap manual replacement of a disk without rebooting the system (available only for systems with a backplane that supports hot-swapping).

For example if a disk fails in RAID 10, the array remains functional and data is read from the surviving mirrored disk(s). Up to two disk failures can be sustained, depending on how the mirrored pairs fail. Mirror rebuilding A broken mirror can be rebuilt after a new disk is inserted and the

disk is designated as a spare. The system does not have to be rebooted.

Mirror splitting A mirrored array can be split, so that the volumes can be used independently.

VST. See Split Mirrored Arrays (supported only by RAIDCore VST) on page 90.

Multiple RAID levels per disk Support for multiple array levels per disk allows the administrator to create arrays of different RAID levels using the same disks. Native Command Queuing (NCQ) Native Command Queuing is a command protocol of disks that are

supported by RAIDCore. NCQ enables individual disks to internally optimize the order in which Read and Write commands are executed. RAIDCore permits a queue depth of up to 32 read/write commands per disk.

Online Capacity Expansion (OCE) OCE is a process that allows the user to add storage capacity to an existing array, without taking the system offline. OCE enables the user to increase the total storage capacity of an array by integrating unused storage into the array.

Data can be accessed while the disks are added and while data on the array is being redistributed.

VST. See Expanding Disk Capacity Online: Using OCE (supported only by RAIDCore VST) on page 26.

(21)

Online RAID Level Migration (ORLM)

With online RAID level migration, users can easily move an array from one RAID level to another. While the migration is taking place, data is accessible and protected to the lowest protection of either the source RAID level or the destination RAID level.

VST. See Migrating RAID Levels Online: Using ORLM (supported only by RAIDCore VST) on page 26.

RAID Level Support Supports RAID levels 0, 1, 1n, 5, 10, 10n, 50, and Volume.

NOTE: RAID 5 is supported by RAIDCore Plus and

RAIDCore VST. RAID 1n, 10n, and 50 are supported by RAIDCore VST.

Self-Monitoring Analysis and Reporting Technology (SMART)

The SMART feature monitors disks and attempts to detect and report potential or impending failures. Its purpose is to be proactive: to allow the user to replace a disk before it degrades or fails.

This feature helps monitor disk performance and reliability, and protects the data on the disk. When problems are detected on a disk, the user can replace or repair the disk without losing any data. SMART-compliant disks have attributes for which data (attributes) can be monitored, in order to identify changes in values and determine whether the values are within threshold limits. Many mechanical and electrical failures display some degradation in performance before a failure occurs.

Software RAID spanning A feature of the operating system, software RAID spanning allows the creation of arrays larger than 2 TB. Because Windows has a limit of 2TB of space per array, see Table 51 on page 112 to create volumes larger than 2TB.

(22)

RAIDCore Licensing Overview

RAIDCore ships with a 30-day trial license that can be activated only once on any given system and expires 30 days after the activation. The trial license is included with the RAIDCore installation and does not need an Internet connection to be activated.

Major Features of RAIDCore Licensing

• Programmatic Internet-based online activation: This is done via connection to the Dot Hill Systems Corp. License server at install time or later (requires an Activation ID).

NOTE: Activation can only be done once for a given client system and Activation ID. Any subsequent

activation request involves Dot Hill Systems Corp. technical support. Minor license issues due to hardware changes can be handled via repair (see below). Activation can also be completed after the installation through the GUI or command line tools.

• License event notifications: Uses standard RAIDCore event mechanisms (bc_service.log, bc_popup, GUI, e-mail if configured), consistent with other RAIDCore events.

• License validation check: A license validation check is done to ensure a valid license. A successful license check results in a heartbeat message to the RAIDCore driver.

NOTE: RAIDCore Event Service <bc_service.exe> must be running during the license validation check.

• RAIDCore heartbeat verification checks every hour to make sure that the services are running. If the heartbeat message is missing, the driver generates a warning event after an hour, and reduces performance after another hour. Performance is restored immediately if a valid license check message is received. • License validation check will cause a significant reduction in RAIDCore performance after a grace period,

if the check fails or shows an expired license. The grace period varies depending on circumstances. • RAIDCore event notification will send a periodic reminder during the grace period (once a day) and more

frequent notification after the grace period has elapsed (twice a day).

(23)

Features Set: by RAIDCore License Level

Table 4 Feature Set for RAIDCore: by RAIDCore License Level

License Levels Features Sub-Features RAIDCore

Basic

RAIDCore Plus

RAIDCore VST

Legacy/boot drive support Yes Yes Yes

O-ROM (BIOS) supporta

Create array Yes Yes Yes

Delete array Yes Yes Yes

Boot/INT13 control Yes Yes Yes

RAID levels supportedb

Volume Yes Yes Yes

0 Yes Yes Yes

1 Yes Yes Yes

10 Yes Yes Yes

5 No Yes Yes

1n No No Yes

10n No No Yes

50 No No Yes

Array creation

No initialization Yes Yes Yes

Foreground

initialization Yes Yes Yes

Background

initialization Yes Yes Yes

Array copying No No Yes

Array deletion Yes Yes Yes

Mirror splitting No No Yes

Global Yes Yes Yes

Sparing Dedicated Yes Yes Yes

Distributed No No Yes

Consistency check Background Yes Yes Yes

Scheduled Yes Yes Yes

Online capacity expansion (OCE)c No No Yes

Online RAID level migration (ORLM) No No Yes

Drive/disk roamingd

Same-system support Yes Yes Yes

Between-systems

support Yes Yes Yes

(24)

a. At the BIOS Configuration Utility (also referred to as the option-ROM or O-ROM) arrays can be created or deleted, and Critical or Offline arrays are indicated. INT13 support can be turned off completely at the BIOS Configuration Utility.

b. See Chapter 3 Arrays, Disks, and RAID Levels for detailed information about understanding arrays, RAID levels, and performance and

reliability considerations.

c. The unique ability of RAIDCore to provide online expansion to RAID levels across multiple disks becomes extremely valuable when expanding storage is a requirement.

d. Disk roaming allows arrays to be moved from port to port, either within the same system or between systems.

e. Multiple RAID levels (array types) per disk allows the administrator to create different RAID levels on the same disks. For example: The administrator wants data redundancy for the user data, and creates a RAID 5 set using part of the disks’ data. At the same time the administrator wants performance for the swap spaces, and creates a RAID 0 array using the rest of the disks’ capacities (space). This feature is useful in collecting unused capacity from disks with different capacities.

f. This feature increases data integrity for redundant array types, by logging areas of an array that have been written to. In the event of a system crash, the logged area’s consistency is checked and/or corrected. Without this feature, data corruption might occur.

g. Arrays can be instantly created and used by skipping the background consistency check. For certain types of redundant arrays this is a viable option and has no data integrity drawbacks. A consistency check can always be done at a later time. If an initialization is skipped when using RAID 5 or RAID 50, the array will not be redundant until a consistency check is performed.

h. Disks can be added to the system and to an array while the system is operating.

Online license activation Yes Yes Yes

Drive interfaces supported

SATA Yes Yes Yes

SAS Yes Yes Yes

SATA CD/DVD Yes Yes Yes

ATAPI Yes Yes Yes

Dissimilar disk support within the same

array Yes Yes Yes

Cache support

No Cache Yes Yes Yes

Read Cache Yes Yes Yes

Write Back Cache Yes Yes Yes

Read with Write

Back Cache Yes Yes Yes

Create Array and Delete Array functions

without rebooting Yes Yes Yes

Restore (rebuild) priority Yes Yes Yes

Multiple RAID levels per diske Yes Yes Yes

Touched region loggingf Yes Yes Yes

E-mail event notification Yes Yes Yes

System event log integration Yes Yes Yes

Instant create supportg Yes Yes Yes

Hot-swap supporth Yes Yes Yes

Table 4 Feature Set for RAIDCore: by RAIDCore License Level (continued)

License Levels Features Sub-Features RAIDCore

Basic

RAIDCore Plus

(25)

3 Chapter 3 Arrays, Disks, and RAID Levels

Understanding Arrays

Arrays are several disks that are grouped together to improve either the performance or reliability of a storage system. Because some RAID levels enhance performance while others improve reliability, it is important to consider the user’s needs when planning an array configuration.

RAIDCore supports various RAID levels. See the following table.

NOTE: It is highly recommended that this user manual be reviewed in its entirety before configuring arrays.

Some of the advanced features of RAIDCore (such as sparing options) must be understood by the user before creating arrays.

RAID Levels

RAIDCore supports the RAID levels indicated in the following table.

Table 5 RAID Levels and Selection Criteria

Number of Disks RAID Level

1 to 4 Volume, RAID 0, RAID 1, RAID 1n, RAID 5

4 and up RAID 10, RAID 10n, RAID 50

Table 6 RAID Levels - General Characteristics

RAID Level Main Characteristic Use/Usefulness

Volume (JBOD) RAIDCore treats one or more disks or the unused space on a disk as a single array.

1. Concatenation provides the ability to link-together storage from one or several disks, regardless of the size of the space on those disks.

2. Concatenation is useful in scavenging space on disks unused by other disks in the array.

3. Concatenation does not provide performance benefits or data redundancy.

4. When a disk in a concatenated or spanned array fails, data is lost in the array. Because there is no redundancy, data can be restored only from a backup.

RAID 0 (striping) Provides the highest performance but no data redundancy. Data in the array is striped (distributed) across several disks.

RAID 0 arrays are useful for holding information, such as the operating system paging file, where performance is extremely important but redundancy is not. RAID 1 (mirroring) Mirrors data on a partition of one

disk to another.

(26)

24 Chapter 3 Arrays, Disks, and RAID Levels

RAID 1n (n-way mirroring) Mirrors the data stored in one disk to several disks.

1. This RAID level provides superior data redundancy because there are three or more copies of the data.

2. Useful for creating exact copies of an array for backup purposes. 3. Expensive, in both performance and the amount of disk space necessary to create the array type. RAID 10 (striped RAID 1 sets) Combines mirrors and stripe sets.

RAID 10 allows multiple disk failures, up to 1 failure in each mirror that has been striped.

1. Offers better performance than a simple mirror because of the extra disks.

2. Requires twice the disk space of RAID 0 to offer redundancy. RAID 10n Stripes multiple n-way mirror sets.

Allows multiple disk failures per mirror set, up to n-1 failures in each mirror set that has been striped (where n = the number of disks in each mirror set)

1. This RAID level is useful in creating exact copies of an array's data using the Split command. 2. Offers better random read performance than a RAID 10 array, but uses more disk space.

RAID 5 (stripe with parity) Stripes data, as well as parity, across all disks in the array. Parity information is interspersed across the disk array.

In the event of a failure,

RAIDCore can restore the lost data of the failed disk from the other surviving disks.

1. Offers exceptional read performance, as well as redundancy.

2. Write performance is not an issue due to the tendency of operating systems to perform many more reads than writes. 3. Requires only one extra disk to offer redundancy.

4. For most systems with three or more disks this is the correct choice for a RAID level. RAID 50 (striped RAID 5 sets) Intersperses parity information

across each RAID 5 set in the array.

1. Offers good read performance, as well as redundancy. A 6-disk array provides two striped, 3-disk RAID 5 sets.

2. Useful in very large arrays (arrays with 10 or more disks). 3. Can handle multiple disk failures.

Table 6 RAID Levels - General Characteristics (continued)

(27)

Array States

Within the management applications, an array is a logical device that can exist in one of three states: Normal, Critical, or Offline.

• In RAIDConsole, these states display in the Array List section in a column named State. • Within the bcadm Command Line tool, these states also display in a column named State. The array states are defined in the following table.

Whether an array is marked as Critical or Offline depends upon what RAID level it is and how many disks within the array have failed. Note the changes in state in the following table.

For example, RAID 50 is a stripe of RAID 5 sets. If a RAID 50 set contains three RAID 5 sets, each RAID 5 set can have up to one disk failure. The RAID 50 array state shows as Critical. If a third disk fails, the state of the RAID 50 array changes to Offline. Or, if two disks fail on one of the RAID 5 sets, the state of the RAID 50 array changes to Offline.

More than one array can be created using the same set of disks. If a disk is disconnected that belongs to more than one array, only the arrays that try to access the disk and receive I/O errors report the failure. For example: there are two arrays, both of which are RAID 5 sets, and both use disk 4. If a system being used by array 1 receives an I/O error when trying to communicate with disk 4, the state of array 1 changes to Critical. However, the state of array 2 using disk 4 does not change to Critical until an I/O error is reported. If systems using array 1 are not communicating with failed disk 4, the state of array 1 still displays as Normal.

If a rescan of all channels is performed after disconnecting a disk, the state of every array using the missing disk changes from Normal to either the Critical or Offline, depending on the RAID level.

Table 7 Array States

State Description

Normal The Normal state is displayed when everything is functioning correctly.

Critical The Critical state is displayed when the array is no longer redundant (fault tolerant) because of one or more disk failures. Arrays can still be read and written to, but the data is no longer protected should another disk fail.

Offline The Offline state is displayed when arrays cannot be read or written to because of one or more disk failures.

Table 8 Failure States by RAID Level

RAID Level This Failure State... Is Displayed Whenever...

RAID 1, RAID 5 Critical A single disk fails.

RAID 1n, RAID 10,

RAID 10n, or RAID 50 (RAID levels with multiple redundancies)

Critical A single disk fails in any one of those sets.

Volume and RAID 0 Offline A single disk fails.

RAID 1 and RAID 5 (redundant arrays)

Offline Two or more disks fail. RAID 1n, RAID 10, RAID 10n

(RAID levels with multiple redundancies)

Offline All disks in a set fail.

RAID 50

(RAID levels with multiple redundancies)

(28)

Creating Arrays: Future Expansion

When creating arrays, consider whether disk capacity will need to expand in the future. If the file system must be expanded, perform the tasks indicated in the following table.

Expanding Disk Capacity Online: Using OCE (supported only by RAIDCore

VST)

Online Capacity Expansion (OCE) allows: • Adding disks to an array at any time.

• Accessing the array data while it is being redistributed.

To increase the size and organization of an array, transform the array. For more information on transforming arrays, see Transform Arrays (supported only by RAIDCore VST) on page 78.

Migrating RAID Levels Online: Using ORLM (supported only by RAIDCore

VST)

Online RAID Level Migration (ORLM) allows an array to move from one RAID level to almost any other RAID level. This task includes migrating the array from a non-redundant RAID level to a redundant RAID level.

Prior to starting a RAID level migration/transformation, make sure that the disks selected for the destination array have sufficient capacity. RAID level migration/transformation can occur only when the destination array has the same or larger capacity as the source array.

While the migration/transformation is taking place, data is accessible and protected to the lowest protection of either the source RAID level or the destination RAID level.

The Transform task can also be used to expand the capacity of an array, by using OCE. It can also be used as part of the system backup and recovery strategy through the use of the RAID 1, RAID 10, RAID 1n, and RAID 10n RAID levels.

To perform this process, see Transform Arrays (supported only by RAIDCore VST) on page 78.

Table 9 Array Expansion Considerations

For This Operating System Do This... And Consider This...

Microsoft Windows® Format the arrays with NTFS. Microsoft Corporation provides a utility (Diskpart.exe) that can dynamically extend an NTFS file system onto any unused adjacent space.

Note also that using a single partition per array makes expansion much easier.

1. The Diskpart.exe utility version depends on which version of the Windows operating system is running. 2. The Diskpart.exe utility can be found on the CD for some versions of

Windows operating systems, or on the Microsoft Corporation website (http://www.microsoft.com) for other versions. Use the correct version for the operating system. Linux Use an expandable file system. Because RAIDCore software is limited

(29)

Array Tasks: Starting and Stopping Tasks

Tasks are started when one of the following actions are performed: • Create a redundant array.

• Transform an array. • Copy an array. • Restore an array.

• Check for consistency on redundant arrays.

• Verify that data was not corrupted after a system crash (Check_Bitmap; performed automatically).

Full task control can be used on Create, Copy, Consistency Check and Bitmap Check tasks. On a Transform or Restore task for dedicated and global spares, task control can only pause/resume, but it cannot remove the task. To remove these types of tasks, pause and then delete them.

NOTE: If a task is paused and then deleted, the array will be deleted. Deleting a task is the same as deleting

an array. Data loss will occur. Backup all data prior to deleting a task that involves a Transform or a Restore. The tasks indicated in the following table can be displayed for each array.

Table 10 Types of Tasks per Array

Task When Displayed

Transform While an array is being transformed. Create While an array is being created. Copy While an array is being copied.

In a --verbose query, the array numbers of the source and destination array are given (for example, Copy 2>5).

Consistency Check While verifying that the parity (RAID 5) or mirror disk consistency is correct. For redundant type arrays only.

Restore While an array is being restored.

Check_Bitmap While verifying that the parity on a RAID 5 set, or the mirror halves on a RAID 1 or RAID 10 set, are consistent. This action is performed automatically to ensure that data is not corrupted whenever a system crashes.

(30)

Understanding Disks

Disks States

Within the management applications, a disk can be part of one or more arrays and can exist in one of three states: Online, Offline, or SMART Error.

• In RAIDConsole, these states are displayed in the Disk List section in a column named State. See Table 28 on page 70 for additional information.

• Within the bcadm program, these states are also displayed in a column named State. See Table 59 on page 121 and Table 60 on page 122 for additional information.

The disk states are defined in the following table.

If a disk fails it is displayed in the Disk List as Online until a rescan is performed. (If host I/O is going to a failed disk and the disk is used in a single array, the disk appears as Online in the Disk List but is highlighted in red; the disk appears as Failed in Array View. If host I/O is going to a failed disk and the disk is used in multiple arrays, the disk may appear in the Disk List as Offline and as Failed for each array in Array View; after a rescan the Disk List state may change to Online). After the rescan is performed the following can occur: • A disconnected disk no longer appears in the Disk List (although it appears as Missing in the Array View, for

the array(s) that uses the disk).

• A disk that is disconnected, but for which a rescan is not performed, appears in the Disk List as Online. However, if a Background Array Scan or Consistency Check is performed, the disk state changes to Offline and it is highlighted in the Disk List in red.

• A disk that experiences a catastrophic failure appears in the Disk List as Offline and is highlighted in red. • A disk that has a SMART error appears in the Disk List as SMART Error. (A disk with a SMART error can be

used to create an array, but the array must be created in the BIOS Configuration Utility, not in RAIDConsole). • A disk that experiences a software-related failure appears in the Disk List as Online and is highlighted in red.

New arrays can be created with the disk.

Arrays that exist on a failed or disconnected disk might not be designated as Failed or Missing until the system attempts to communicate with the failed or disconnected disk.

RAID Minimum Disk Configurations

The following table lists the minimum disk configurations for each RAID level.

Table 11 Disk States

Disk State When Displayed

Online Whenever the disk is connected, functioning correctly, and RAIDCore can communicate with it.

Offline Whenever the disk fails and RAIDCore detects an error condition on the disk. SMART Error Whenever the disk reports a SMART error(s) to RAIDCore.

Table 12 Minimum Disk Configurations per RAID Level

RAID Level Minimum Number of Disks

(31)

Rescanning Disks for Changes in State

The information displayed in the Disk List section is the state of the disks when they were last scanned. If a rescan has not been performed, the information being displayed is the state of the disks at boot time. Every time a disk is connected or disconnected while online, a message asks if the user wants to perform a rescan (of all SATA channels). If Rescan is selected, the information in both the Array List and the Disk List is updated. This view might show arrays as being in a Critical or Offline state, if all disks have not been installed or removed.

Although it is highly recommended that the system be shut down before adding or removing disks, disks can be added or removed while the system is online (“hot-swapping”), if the system supports the hot-swapping function.

Because of this function, RAIDConsole does not automatically perform a rescan when it detects that a disk has been added or removed. For example, to hot-swap a RAID 5 set with six disks into a new system, do not perform a rescan until all six disks have been connected.

Arrays associated with the disks that are not yet connected change state to either Critical or Offline. In the example above, if the state of the RAID 5 set changes to Offline, data is lost.

This feature can also result in the state of a disk being reported differently in the Array View and the Disk List. A disk within an array can have a state of Failed in the Array View field, while at the same time it can show a state of Online in the Disk List.

Sparing Options: Disks and Arrays

RAIDCore supports multiple sparing options. Spares are restored in the order indicated in the following table.

NOTE: An array is marked Critical or Offline if a disk reports a Failed state to an I/O, or if the SAS/SATA

cable or power cable is disconnected.

NOTE: One or more spares can be assigned to a redundant RAID level.

NOTE: Spare assignments do not apply to non-redundant RAID levels. To protect data, transform the array

to a redundant RAID level. Spares can then be assigned. (Be aware that the Transform task is supported only by RAIDCore VST.)

Table 13 Sparing Options

Option Description

Dedicated A spare disk assigned to a specific, redundant array.

(32)

Distributed Sparing (supported only by RAIDCore VST)

RAIDCore’s proprietary distributed sparing feature reserves space on each disk in an array. This space is used when the data from a failed disk is being regenerated during a Restore task.

Whereas other methods of sparing (such as dedicated spares) provide the same level of protection, distributed sparing provides better performance. All disks are active in an array and are not sitting idle, as is the case with dedicated spares.

Another advantage of distributed spares is that because all disks are active, a disk cannot fail and go unnoticed, as is the case with dedicated or global spares.

With distributed sparing, each array has its own dedicated fail-over spaces. This averts the potential problem of having insufficient space to start a fail-over on the single disk that has been assigned as a spare.

A distributed spare is assigned at the time an array is created or transformed. Distributed spares are valid only for:

• RAID 5 (four or more disks)

• RAID 50 (four or more disks per RAID 5 set) • RAID 10 and RAID 10n (six or more disks).

This spare type is the most protective, because space is allocated when the array is created. Like a dedicated spare, this spare type is assigned to a specific array.

If an array was created without a distributed spare, the array must be transformed to add a distributed spare. A distributed spare can be assigned only when an array is being created or transformed.

• If there is insufficient unused space on the disks in an array, a distributed spare cannot be added without adding an additional disk and transforming it to an array (with distributed sparing enabled).

• The distributed spare option uses the equivalent storage of one of the disks in the array.

For example: the total capacity of six disks is being used in a RAID 5 array, and distributed sparing has been enabled.

• The capacity of the array is the same as a four-disk RAID 0. • The capacity of one disk is lost to the RAID 5 parity data.

• The capacity of another disk is lost to the distributed sparing option.

For additional information, see Add or Remove Distributed Spares (supported only by RAIDCore VST) on page 92.

Dedicated Sparing

A dedicated spare is a disk that is assigned as an alternate disk for a specific array. Should a disk fail in that array, the alternate disk is used to replace the failed disk and the array is rebuilt.

A dedicated spare can be assigned to any redundant array type, and up to four spares can be assigned to an array.

NOTE: Assigning a dedicated spare does not reserve space on the disk. Therefore, an automatic restore is not

guaranteed if a disk fails. If a disk fails, make room on the disk for the fail-over to complete, or assign a different disk with enough room. If a dedicated spare is assigned and a disk fails, the restore process starts automatically, if there is enough space available on the dedicated spare.

NOTE: A dedicated spare cannot be assigned while a task is running on the array.

NOTE: When a dedicated spare is assigned while an array is being created or transformed, the first disk

(33)

NOTE: When a dedicated spare is added after the array has been created, select which disk to use as the

spare.

For additional information, see Add or Remove Dedicated Spares on page 92.

Global Sparing

A global spare is a disk that is assigned as an alternate disk for multiple arrays, instead of associating it with only one array.

Many arrays can be restored using the global spare disk, as long as it is not already part of the array and it has enough space available. Unlike a dedicated spare, this type of spare can be assigned at anytime, even while tasks are running on arrays.

Assigning a disk for use as a global spare does not reserve space on that disk. An automatic restore is not guaranteed if a disk fails.

If there is not enough disk space on the global spare, make room for the fail-over to complete, or assign a different disk with enough capacity as the spare. If there is enough space available on the global spare and a disk failure occurs, the restore process starts automatically.

For additional information, see Add or Remove Global Spares on page 93.

RAID Performance Considerations

With RAID technology, performance is based on the following considerations: • The number and organization of disks in an array.

• Caching attributes used for the array. • Application workload.

Number and Organization of Disks

RAID functions increase performance by putting more disks to work and by buffering data for the host. Many disks can transfer data at greater than 50 MB per second. RAIDCore can aggregate this bandwidth in an almost linear fashion, as more of the same disks are included in an array.

Caching Attributes

RAIDCore can also be configured to provide read and Write Back caching, if desired. Write Back caching has a large effect on most workloads, but should be used with caution.

Application Workload

(34)

RAID Reliability Considerations

RAID reliability is enhanced through data redundancy and backup.

Data Redundancy

RAID 1, RAID 1n, RAID 5, RAID 50, RAID 10, or RAID 10n are necessary for redundancy. With redundancy, both capacity and performance are sacrificed for reliability. With RAIDCore, extremely high performance is obtained even with redundant-type arrays.

Backup

NOTE: The Transform and Mirror Splitting tasks are supported only by RAIDCore VST.

RAIDCore’s ability to transform arrays and split mirrors can be used as part of a backup strategy.

For example, create a RAID 1 or RAID 10 array, yet periodically transform them into RAID 1n and RAID 10n arrays. The RAID 1n and RAID 10n arrays can be split into the original array and a backup array.

As part of a “hot fallback” strategy, the backup array can be kept online and hidden from the operating system or remain visible, or it can be removed from the system and stored as a backup device.

When a boot array is split, the original and the copy are identical and contain the same labels. It is

recommended that you keep the split copy hidden during boot time, to avoid boot-time issues with duplicate labels.

Flexibility and Expansion Considerations

Before configuring an array, consider the following points to enhance the flexibility of the RAID system.

Multiple RAID Levels

With RAIDCore, different RAID levels can be created on the same disk, to adapt each array to the I/O that it processes. Also, more than one array can be created per disk.

Depending on the array capacity and redundancy level, an existing array can be transformed to another RAID level, if the level of the array being used is not the optimal RAID level for the application. Also, different arrays with different characteristics can be built for different applications.

Operating System Considerations

An important array planning consideration is whether to extend the RAID benefits to the system disk and operating system, as well as to the data disks.

For example, if the operating system is installed on a RAID 1 array (mirror), split the array into two volumes. Hide one of the resulting volumes. The operating system can be recovered easily if the system disk fails. Alternatively, improve operating system performance by installing the operating system on a RAID 10 array. Split the array and hide one of the RAID 1 arrays.

(35)

4 Chapter 4 Installing Software

Before You Begin...

This chapter is designed to get RAIDCore up and running by using default settings. For most users these are the optimal settings.

CAUTION: All components (drivers and applications) must be installed when installing or upgrading

RAIDCore. This means that a previous version driver will not work with the current version of RAIDCore application, and vice versa.

NOTE: To modify a default setting, see the applicable features in Chapter 5 BIOS Configuration Utility, and

either Chapter 6 RAIDConsole Graphical User Interface (GUI) or Appendix B bcadm Command Line Tool.

System Setup Process

A generic system setup process is described in the following table.

Table 14 System Setup Process

Process Action

1. Startup the system. 2. Access the BIOS setup window for the system.

Set the SATA mode per the type of chipset. For supported AMD chipsets, set the SATA mode as RAID (for use with RAIDCore).

3. Enable the applicable chipset.

Refer to the BIOS configuration procedure in the motherboard’s instruction manual.

4. Initialize the disks. See Initialize Disks on page 54. 5. Create arrays using the

RAIDCore BIOS Configuration Utility.

See Create Arrays on page 55.

6. Copy the drivers to removable storage media.

See Copy Drivers to Removable Storage Media: Windows Operating Systems on page 34.

or

Copy Drivers to Removable Storage Media: Linux Operating Systems on page 35.

7. Install the drivers on the system.

See Install Drivers on a System at the Same Time That the Windows Operating System is Installed on page 36.

or

Install Drivers on a System at the Same Time That the Linux Operating System Is Installed on page 41.

8. Install the Management Suite on the system.

See Install Drivers, Configuration Devices, and the Management Suite: Windows on page 37.

or

(36)

34 Chapter 4 Installing Software

Copy Drivers to Removable Storage Media: Windows Operating

Systems

When installing the operating system to a RAIDCore bootable array, copy the drivers for RAIDCore to removable storage media (CD-ROM, DVD, flash drive, floppy disk, etc).

Use a system that is running a Windows operating system and has a a CD-ROM or DVD drive, floppy drive, or an I/O port for removable storage media, such as a flash drive.

1. Insert the storage medium into the applicable drive, or connect a flash drive to a USB or similar port.

• Windows Server 2003: Insert a floppy disk into the disk drive.

• Windows Server 2008 or Windows Server 2008 R2: Insert a blank CD-ROM or DVD into the applicable drive, or connect a flash drive to a USB or similar port.

2. Insert the RAIDCore Media CD-ROM into a second CD-ROM or DVD drive.

3. Copy the files from the applicable folder on the RAIDCore Media CD-ROM, as indicated in the following

table.

NOTE: If the drivers are being installed on a system that has an existing operating system, skip step 4 and

proceed to Install Drivers, Configuration Devices, and the Management Suite: Windows on page 37.

Use the storage medium and its copied files to install the drivers. See Install Drivers on a System at the Same Time That the Windows Operating System is Installed on page 36.

Table 15 Driver Files for Windows Operating Systems

Operating System Files to Copy to a Floppy Disk or Storage Media

Windows Server 2003 SP2

For 32-bit Windows Server 2003 systems: copy all of the files in the windows\driver\bcraid\i386\W2k3 folder to the floppy disk. For 64-bit Windows Server 2003 systems: copy all of the files in the windows\driver\bcraid\x86_64\W2k3 folder to the floppy disk.

Windows Server 2008 SP2, Windows Server 2008 R2

(37)

Copy Drivers to Removable Storage Media: Linux Operating Systems

To install a bootable array, copy the drivers on the RAIDCore Media CD-ROM to removable USB storage media.

• A SATA DVD drive cannot be used to load the drivers.

• A USB DVD drive can be used to install the operating system and drivers.

Removable storage media are needed only if a bootable array is installed. If the installation is not to a bootable array, follow the procedure Install Drivers, Configuration Devices, and the Management Suite: Linux on page 49.

1. Connect a flash drive to a USB port.

2. Insert the RAIDCore Media CD-ROM into the system’s CD-ROM or DVD drive.

3. Mount the RAIDCore Media CD-ROM at a convenient location. (The RAIDCore Media CD-ROM is in

ISO 9660 format).

Example: mount /dev/cdrom /mnt/cdrom

4. To find the kernel version, processor type, and architecture, at the command prompt, type: uname -a

Example output: Linux <system>.<domain> 2.4.21-4.ELsmp #1 SMP <date & time> i686 GNU/Linux

5. From the <cd mount point>/linux/<vendor>/<version>/driver folder, select the

distribution version of Linux.

6. Copy the .iso image file (which match the distribution of Linux) to the desktop. • For Red Hat Linux

• Copy the .iso file from the desktop to the USB flash drive.

• For SUSE Linux

• Mount the .iso image file. For example: <mount -o loop “the name of the iso file”/media>

• Navigate to /media and copy the .iso folder/files from /media to the USB flash drive.

NOTE: If the system/server used to copy the .iso image to the USB flash drive uses a Windows operating

system, perform “a” or “b” below:

a. Use an application (such as WinImage) to extract the .iso image directly onto the USB flash drive. or

b. Use an application (such as Nero Express Essentials) to burn the .iso image onto a CD-ROM. Then, copy the burned .iso image from the CD-ROM onto a USB flash drive.

(38)

36 Chapter 4 Installing Software

Install Drivers and Applications: Windows Operating Systems

There are two situations under which RAIDCore can be installed:

• At the same time that the Windows operating system is installed. See Install Drivers on a System at the Same Time That the Windows Operating System is Installed below.

• After the Windows operating system has been installed. See Install Drivers, Configuration Devices, and the Management Suite: Windows on page 37.

Install Drivers on a System at the Same Time That the Windows Operating

System is Installed

CAUTION: All components (drivers and applications) must be installed when installing or upgrading

RAIDCore. This means that a previous version driver will not work with the current version of RAIDCore applications and vice versa.

Follow these functions, if the operating system is being loaded onto a RAIDCore bootable array: • Modify the boot priority list in the motherboard BIOS. See the motherboard manual for information. • Make sure that the RAIDCore bootable array is included in the boot priority list.

• At the BIOS Configuration Utility, make sure that INT13 support is ON for RAIDCore. See Change the Controller Options on page 60.

• At the BIOS Configuration Utility, make sure that the boot array is the first array listed in the Arrays list. Use the Swap Two Arrays task to place the boot array first. See Swap Arrays on page 58.

Install the drivers

1. Follow Microsoft procedures to install the applicable Windows operating system.

NOTE: At step 2 for Windows Server 2003, press F6 immediately. 2. At the Windows Setup window, perform the following:

• Windows Server 2003: Press F6 to install the driver.

• Windows Server 2008 or Windows Server 2008 R2: Click Load Driver to install the driver.

3. When prompted, insert the storage medium into the applicable drive or USB or similar port:

• Windows Server 2003: Press S, then press Enter.

• Windows Server 2008 or Windows Server 2008 R2:Press Load Driver, then press Enter.

4. Windows should report that it found the drivers. Press Enter to select and load the drivers.

5. Press Enter to continue. Follow the on-screen instructions to complete the installation of the applicable

Windows operating system.

6. After the operating system is installed, see Install Drivers, Configuration Devices, and the Management Suite:

Windows on page 37.

RAIDCore User Manual P/N Revision A December 2009

Contents

Figures

Tables

Preface

Intended audience

Prerequisites

Document conventions and symbols

Customer Support

1

Chapter 1 Safety Precautions

General

Safety Definitions

Safety Precautions

2

Chapter 2 Getting Started

RAIDCore Technology

Who Should Use This Manual

System Requirements for Using RAIDCore

Supported Controllers

Supported Operating Systems

Features of RAIDCore

RAIDCore Licensing Overview

Major Features of RAIDCore Licensing

Features Set: by RAIDCore License Level

3

Chapter 3 Arrays, Disks, and RAID Levels

Understanding Arrays

RAID Levels

Array States

Creating Arrays: Future Expansion

Expanding Disk Capacity Online: Using OCE (supported only by RAIDCore

VST)

Migrating RAID Levels Online: Using ORLM (supported only by RAIDCore

VST)

Array Tasks: Starting and Stopping Tasks

Understanding Disks

Disks States

RAID Minimum Disk Configurations

Rescanning Disks for Changes in State

Sparing Options: Disks and Arrays

Distributed Sparing (supported only by RAIDCore VST)

Dedicated Sparing

Global Sparing

RAID Performance Considerations

Number and Organization of Disks

Caching Attributes

Application Workload

RAID Reliability Considerations

Data Redundancy

Backup

Flexibility and Expansion Considerations

Multiple RAID Levels

Operating System Considerations

4

Chapter 4 Installing Software

Before You Begin...

System Setup Process

Copy Drivers to Removable Storage Media: Windows Operating

Systems

Copy Drivers to Removable Storage Media: Linux Operating Systems

Install Drivers and Applications: Windows Operating Systems

Install Drivers on a System at the Same Time That the Windows Operating

System is Installed

Install the drivers