This chapter describes the Array Manager storage model and its components. It also covers concepts associated with the redundant RAID Fibre Channel controllers. In addition, the chapter gives information about what RAID is and how different levels of RAID work. The chapter's topics are:
The Array Manager storage model represents the different components in a storage subsystem as either physical or logical storage objects. The storage objects are in a hierarchical order. A computer object representing a local or remote computer running Array Manager is at the top of the hierarchy. You can have multiple computers at this level. Each computer displays a default tree view, which you can change. By default, the tree view displays three main objects below each computer:
In the sample screen above, the computer object at the top of the hierarchy, DELL4, represents the local computer running Array Manager. It has three storage objects beneath it, Arrays, Disks, and Volumes. The plus sign in front of each object indicates that additional subordinate objects exist under that object.
All the storage objects have context menus associated with them. To access a storage object's context menu, right-click on the object.
Arrays represent the physical and logical storage subsystems connected to RAID controllers. The array consists of one or more subsystem objects. An example of a subsystem can be a PERC subsystem, which is a RAID controller-specific storage subsystem. A subsystem object consists of two subordinate storage objects:
Physical array — Contains the physical view of storage entities such as controllers, channels, and array disks, as well as physical enclosures and SAF-TE objects.
Logical array — Contains the logical view of array groups. An array group contains array disks and virtual disks.
In the sample screen above, the fully expanded Arrays object hierarchy shows both the Physical Array and the Logical Array. The Physical Array object has a PERC 2/SC controller that contains a single channel, Channel 0, with five array disks. The array disk numbering corresponds with the number of the channel and the SCSI ID. In the screen above, the array disks 0:0, 0:1, 0:2, 0:3, and 0:4 are connected to Channel 0 and occupy SCSI IDs 0 to 4.
The Logical Array object contains array groups. An array group has array disks that are controlled by a particular array controller. An array group is named by the number of the array controller it is associated with. For example, array disks attached to Controller 0 belong to Array Group 0. With SCSI RAID controllers, you can create multiple virtual disks from disks in an array group. However, with a redundant Fibre Channel RAID controller, which supports a large number of physical disks, you will need to define one or more disk groups from the array group before creating virtual disks. You then create virtual disks from a disk group rather than from the larger array group. Disk groups are not shown in the screen above.
A virtual disk is an abstract entity made up of array disks and/or array disk segments presented to an operating system as a single contiguous block of storage space. When you create a virtual disk, you are asked to specify a hardware RAID level.
An array disk is a disk controlled by the array controller. This disk can be placed in an array group, and if you are using redundant Fibre Channel RAID controllers, in a disk group.
As shown in the sample screen above, virtual disks created in a particular array group are listed under that Array Group object. These virtual disks may also appear as Microsoft Windows NT® or Windows® 2000 disks under the Disks object. Virtual disks are placed under the Disksobject because once a virtual disk is created, it is viewed by the operating system as a regular hard disk.
Note: When you have a PowerVault™ 660F Fibre Channel Controller
that is part of a Storage Area Network (SAN), you will have to first use
the Dell OpenManage Storage Consolidation software to assign the
virtual disk to the server before it will appear under the Disks object.
For Windows NT servers, this may also require a reboot.
Disks represent the disks recognized by the Microsoft Windows NT or Windows 2000 operating system. This can include regular hard disks and virtual disks created through Array Manager. This view also includes removable media, such as CD-ROM drives or removable disks, such as Zip disks. The screen below shows an example of a Disks section of the tree view.
A basic disk adheres to the partition-oriented scheme of Windows NT, Windows 95®, Windows 98®, and MS-DOS®. Basic disks can also contain RAID volumes that were created in NT Disk Administrator, including spanned volumes (volume sets), mirrored volumes (mirror sets), striped volumes (stripe sets), and RAID-5 volumes (stripe sets with parity). In addition, virtual disks, CD-ROMs, and removable-media disks are considered basic disks.
Dynamic disks are created by upgrading basic disks using Array Manager. A dynamic disk is a physical disk that can contain dynamic volumes created by Array Manager.
For more information about disks, see the Disk Management chapter.
A volume is a logical entity that is made up of portions of one or more physical disks. A volume can be formatted with a file system and can be accessed by a drive letter. The maximum size of a volume depends on the quantity of free disk space and the type of volume selected. The screen that follows shows an example of a Volumes section of the tree view.
In Array Manager, primary and extended partitions and logical drives exist only on basic disks. Basic disks use the traditional disk partitioning mechanism used by MS-DOS, Microsoft Windows 95, Windows 98, and Microsoft Windows NT. A basic disk can have up to either four primary partitions or three primary partitions plus an extended partition. This extended partition can be subdivided into as many as 32 logical drives.
A logical drive is a simple volume that resides on an extended partition of a basic disk. You can use all or part of the free space in an extended partition when creating logical drives.
Basic volumes refer to all the volumes that are on basic disks. Basic volumes can be primary or extended partitions, simple logical drives that reside on extended partitions, or RAID volumes that were originally created in Windows NT Disk Administrator.
A dynamic volume is a logical volume that can be created from one or more dynamic disks using Array Manager software. Dynamic volume types include simple, spanned, striped, mirrored, and RAID-5. By using spanned dynamic volumes, you can dynamically increase the size of your dynamic volumes as the need arises.For more information, see Working with Dynamic Volumes in the chapter on Volume Management.
For detailed instructions about creating and/or managing Array Manager storage model objects, see the specific sections as follows:
Redundant Fibre Channel RAID controller configurations interconnect two identical controllers that share a common set of array disks. This configuration allows a surviving controller to take over resources of a failed controller. This failover process is transparent to the applications running on the host.
Redundant Fibre Channel RAID controller support provides the system with the mechanisms for the following:
Initializing two controllers to process host requests
Establishing a controller-controller nexus (C-C nexus) for controller redundancy
Detecting failure of a controller that is part of a C-C nexus and initiating a failover
Detecting the replacement of a failed controller, initiating a failback and re-establishment of a C-C nexus.
Note: A controller-controller nexus refers to the state in which both
redundant controllers are in communication. In this state, each
controller can copy write-back data to its partner controller and can
determine whether the other controller is operating.
The two Fibre Channel RAID controllers in the disk enclosure will both be connected to a server in a direct attach configuration or in a SAN configuration. The two controllers communicate with each other to verify that both are functioning properly through a ping/acknowledgment sequence. Failure to acknowledge the ping triggers failover.
The redundant Fibre Channel RAID controller configuration also supports dual or multiple server communication in a SAN, offering the advantages of being able to sustain data access in the event of a host failure. If configured in a cluster or high-availability environment, it is also able to sustain data access in the event of the failure of a server or a host bus adapter (HBA). This configuration requires implementation of alternate path software on the server.
Both controllers in a redundant Fibre Channel RAID controller system must be identical. The following requirements must be satisfied in order to provide optimal operation.
Both controllers must have the same firmware version and type.
Both controllers must have the same amount of memory.
In redundant Fibre Channel RAID controller configurations, maintaining continuous access to data requires that a failed controller be replaced in a manner that is transparent to the host. Drive channel ports on the controllers are configured as inactive ports until those ports are needed to respond to requests of a failed controller.
In the event of a controller failure in a redundant Fibre Channel RAID controller system, the failed controller's operations are assumed by the surviving controller. The failed controller can then be removed and replaced while the system is online. When the surviving controller detects the presence of the new controller, the new controller resumes processing array operations. During failover and failback, write cache coherency is maintained with the disk drives.
When a controller is participating in a controller-controller nexus (C-C nexus) and detects a communication error with its partner controller, it initiates the failover process. The following steps outline the failover process executed by the surviving controller:
On detection of the controller failure, the surviving controller holds the failed controller in disable partner mode.
The surviving controller activates the failover port.
Cached data is flushed to the disk drives.
Conservative Cache is enabled (if the Conservative Cache Mode controller option has been enabled).
The surviving controller begins handling I/O requests for the failed controller.
When a failed controller is replaced, the system either automatically detects the replacement (if the Auto Restore controller option has been enabled) or is informed of the replacement by the Enable Partner command. The following steps outline the failback process executed by the surviving controller:
A replacement controller is detected.
The surviving controller enables the replacement controller.
Once the replacement controller completes initialization and is ready to resume I/O requests, the surviving controller quiesces both ports by responding with BUSY status to new I/O requests.
The surviving controller disables the failover port.
The surviving controller clears the BUSY condition.
The replacement controller enables its primary ports.
Both controllers disable Conservative Cache mode (if enabled) for write-back system drives and resume normal redundant Fibre Channel RAID controller operation.
A data transfer interface technology that allows for high-speed I/O and networking functionality in a single connectivity technology. The Fibre Channel Standard supports several topologies, including Fibre Channel Point-to-Point, Fibre Channel Fabric (generic switching topology), and Fibre Channel Arbitrated Loop (FC_AL).
The simplest Fibre Channel connection. This provides a direct connection between the transmit output of one node and the receive input of a second node. A second connection is provided between the opposite connectors in order to complete the signals. The physical connection between the two nodes is called a link. No switches, loops, or fabric elements are needed. This is also known as a direct attach configuration.
A topology that uses one or more switches to route frames between nodes in a Fibre Channel network. The routing of frames is transparent to the nodes or devices. This is also known as a Storage Area Network (SAN) configuration.
For more information on Fibre Channel technology, refer to the following web sites:
RAID (Redundant Array of Independent Disks) is a collection of specifications that describe a system for ensuring the reliability and stability of data stored on large disk subsystems.
RAID provides the following general benefits:
Increase of storage system reliability
Improvement of I/O performance
Different RAID levels provide varying degrees of improved reliability and performance. See Choosing RAID Levels for a description of RAID levels and their advantages and disadvantages.
RAID provides various features to ensure high availability of data in the event of drive failure. These features include data duplication and disk drive rebuilding.
Consistency Check — Verifies the accuracy of redundant data in a RAID group. For example, in a system with parity, checking consistency computes the parity of the data drives and compares the results with the contents of the parity drive.
Disk Striping — Storing data in stripes on two or more physical disks. Data is allocated alternately and evenly (in stripes) to the disks of the striped RAID group. Striped RAID groups are not fault tolerant.
Concatenation — Storing data on either one disk (generally considered a simple volume) or on disk space that spans more than one disk (a spanned volume). In Array Manager, a dynamic volume that spans to separate areas of the same disk is also considered concatenated. In Array Manager, a simple volume is defined as a volume that occupies a single contiguous space on one disk.
Disk Mirroring — Duplicating data on two physical disks. It provides data redundancy by using a copy (mirror) of the RAID group to duplicate the information contained in the RAID group. The mirror is located on a different disk. If one of the physical disks fails, the system can continue to operate using the unaffected disk. Both drives contain the same data at all times. Either drive can act as the operational drive. A mirrored RAID group is comparable in performance to a RAID-5 group in read operations but faster in write operations.
Parity — Redundant information that is associated with a block of information. Parity is a calculated value used to reconstruct data after a failure. If a single disk drive fails, it can be rebuilt from the parity and the data on the remaining drives.
Hot Spare — An extra, unused disk drive that is part of an array disk subsystem. A hot spare is always in standby mode. If a disk failure occurs, the hot spare replaces the failed drive without interrupting the system.
Disk Rebuild — A rebuild is the recreating of data that existed on a drive before it failed. You can rebuild data only in arrays with data redundancy, such as RAID level 1. (See Choosing RAID Levels for more information about RAID levels.)
Hot Swapping — A hot swap is the manual replacement of a defective physical disk unit while the computer is still running. When a new drive has been installed, you must issue a command to rebuild the drive.
Fault Tolerance — Includes features like automatic failed drive detection, rebuild, and hot swapping.
RAID defines a collection of different configuration levels (RAID levels), which are used to configure an array of disks. This section describes how RAID levels (RAID-0, RAID-1, RAID-3, RAID-5, RAID-50, RAID-10, RAID-0+1, and RAID-1 Concatenated) provide various levels of performance and redundancy. From these, you can understand what choices are applicable to your particular situation.
Groups n disks as one large virtual disk with a capacity of (n-1) disks.
Redundant information (parity) is alternately stored on all disks.
When a disk fails, the virtual disk still works, but it is operating in a degraded state. The data is reconstructed from the surviving disks, and the parity alternately stored on the disks.
Better read performance, but slower write performance.
Redundancy for protection of data.
Note: Both RAID-3 and RAID-5 have striping with distributed parity.
With RAID-3 the parity is usually dedicated to a single array disk, but
the PowerVault™ 660F RAID controller implements RAID-3 in the
same manner as RAID-5, with distributed parity across multiple disks.
RAID-50 is a concatenation of RAID-5 across more than one three-disk span. For example, a RAID-5 array that is implemented with three drives and then continues on with three more array drives would be a RAID-50 array.
Groups n*s disks as one large virtual disk with a capacity of s*(n-1) disks, where s is the number of spans and n is the number of disks within each span.
Redundant information (parity) is alternately stored on all disks of each RAID-5 span.
Better read performance, but slower write performance.
Requires proportionally as much parity information as standard RAID-5.
Data is concatenated across all spans. RAID-50 is more expensive in terms of disk space.
It is possible to implement RAID-50 even when the hardware does not support it directly. What you can do is to implement more than one RAID-5 virtual disk. Convert the RAID-5 disks to dynamic disks. You then create a dynamic volume that is spanned across all RAID-5 virtual disks.
RAID-10 on PERC 2/SC, 2/DC, 3/DCL, 3/DC, and 3/QC controllers is implemented as RAID-1 Concatenated.
RAID-1 Concatenated is a RAID-1 array that spans across more than a single pair of array disks. This combines the advantages of concatenation with the redundancy of RAID-1. No striping is involved in this RAID type.
Also, RAID-1 Concatenated can be implemented on hardware that supports only RAID-1 by creating multiple RAID-1 virtual disks, upgrading the virtual disks to dynamic disks, and then using spanning to concatenate all of the RAID-1 virtual disks into one large dynamic volume.
This RAID level is used with the PowerVault 660F controller. It is very similar to RAID level 10, but the striping pattern is different. You also can have an odd number of disks.
Groups n disks as one large virtual disk with a capacity of (n/2) disks.
Mirror images of the data are striped across sets of disk drives. This level provides redundancy through mirroring.
When a disk fails, the virtual disk is still functional. The data will be read from the surviving mirrored data stripes.
RAID functions can be implemented with either hardware RAID (a specific RAID controller) or software RAID (software without specific hardware).
Software RAID can be configured on top of the hardware RAID volumes and provide a powerful variety of features. For example, you can mirror a pair of hardware RAID-5 volumes across two RAID controllers to provide RAID controller redundancy.
Note: Careful consideration should be given to a decision between
hardware RAID and software RAID. Software RAID generally has
lower performance; and when it is used with SAN Fibre Channel
storage, the volumes can not be moved from server to server.
