Tài liệu Storage Management pdf

WithWindows 2000 Remote Storage Services RSS and Removable StorageServices RSS, you can automatically move files that are not needed on alocal or network volume to removable media, such

Storage Management

This chapter introduces Windows 2000 storage A new

addition to the operating system is a dynamic storagemanagement environment that essentially provides “free”

fault-tolerance (RAID-1 and RAID-5), enforceable disk quotas,and an MMC plug-in that replaces the clunky disk manage-ment environment of Windows NT 4.0

Windows 2000 Storage

If there are three things that every IT or network tor can be sure of happening regarding storage, they are thefollowing:

administra-✦ No matter how much hard disk storage you plan for, orthink you need, you will always need more

✦ A hard disk will crash or falter within the lifetime of itshost computer system

✦ A hard disk will crash at the worst possible time (usuallyfor you)

Applications and servers crash, and users get peeved whenyou run out of storage or you lose storage When you losestorage that was keeping critical data — data keeping the com-pany alive — your world turns inside out Data loss costs usbillions every year There are many examples where losingdata destroyed a business, and where a good set of backupssaved a business from certain disaster There are four actionsyou need to take to counter or survive the fallout from thethree certainties:

✦ You need to evaluate your storage needs

✦ You need to develop policies for sound storage use

In This Chapter

Overview ofWindows 2000Storage

Storage ManagementDisk ManagementDynamic Volumesand Fault-ToleranceThe Storage QuotaSystem

Troubleshooting

✦ You need to develop and follow a storage implementation plan.

✦ You need to implement a disaster recovery and backup/restore plan

The file system, storage management, and fault-tolerant capabilities of Windows NTwere short on features and manageability, and did not support the 99.9 percent avail-ability initiative (see Chapter 1) Administrators frequently turned to third-partyproducts to keep their data safe Microsoft has responded to your needs, finally, withthree very useful storage services: Disk Management Service, Removable Storage,and Remote Storage These services depend on the NTFS 5.0 file system for many features

In this chapter, you will investigate these three services You will be able to planyour storage needs and decide what components of the Windows 2000 Server stor-age services meet your objectives and needs

Storage Management

To evaluate your storage needs, consider first the practice of storage management

Storage management is practiced on three tiers: You need to manage access to

stor-age (media), you need to manstor-age access to data (availability), and you need to tect data (backup/restore) Let us first look at the access to storage tier.

pro-Access to Storage

Storage is located in two places on a network: on the host server from the tive of the server and on a remote volume from the perspective of clients Computersystems require local storage on which to store system files and boot files to startoperating systems and begin services You do not need large volumes of storage forlocal operating systems and boot files The most any server will require is a partition

perspec-of about 2GB, and that should easily service the OS files and server registry Itshould also be sufficient for local services, processes, upgrades, service packs, uninstall folders, and more

Do the local volumes need to be particularly fast? Not necessarily, because access

to boot files is only needed at startup and a fully functional server, in full tion, seldom updates the system files or the registry So, you do not need to createvolumes that provide fast I/O, and you do not need to spend a lot of money on veryfast disks

produc-Storage considerations for client workstations may be very different than servers,which is what we are largely concentrating on in this chapter However, worksta-tions and thin clients do not need large volumes, and the practice of locating mostapplications and data on network servers (good for licensing, security, and back-ups) is fast becoming the universal practice again

Note

The storage access tier goes down yet another level: data access Within the hostcomputer, you can configure a single disk or an array of disks to hold data Datadisks need to be built according to the needs of your organization and the purpose

of the server

The data volumes will typically hold applications that run on the server, databasesthat hold systems data, such as Active Directory, WINS, DNS and DHCP databases,and more

Your data volumes may need to be fast If your applications or services will makenumerous reads and writes to the data volumes, you will need to plan for fast I/Osolutions Depending on your budget, you may be in a position to spend a littlemore money for fast disks, or you might consider striping volumes instead, or striping volumes on fast disks

You need to note the type of applications that will be opened from the data umes, the type of data that will be accessed from the volumes, and the number ofusers and applications that need to simultaneously read and write to the volumes

vol-You shouldn’t need to calculate data-access time, but certain applications mayrequire you to calculate latency, disk performance, access times, transfer rates, and

so forth Good examples of such applications are busy shopping carts on Web sites

or database servers on the company intranet

The server and hard disks may have to service as many as 1,000 concurrentrequests for data at any given time And these requests may come to the hard disk

as requests to write data and to read data, in which case you need to establishacceptable levels of performance for data access time and for transfer rate (thetime it takes to retrieve and save data to the media)

The three benchmarks of hard disk performance should be understood becausethey may be important to your applications When you are dealing with tens ofthousands of hits a day on a Web site, or a massive data transformation operation,hard disks can take a knock A very busy mail server can chew through severaldisks in its lifetime

The three benchmarks are as follows:

1 Access time This is the time it takes for a hard disk to register a request and

prepare to scan the surface of a disk

2 Seek time This is the time it takes for a hard disk to find and assemble all the

parts of a file

3 Transfer rate This is the time it takes for a hard disk to transfer data on and

off the disk

Another angle on storage access is space If your databases are expanding andusers are generating new files every day, you will need to watch disk space

consumption and plan accordingly Your policy should be one in which every effortshould be made (within budget and the requirements of the business) to continu-ally strive to keep the volumes from filling up more than, for example, 70 percent ofthe available space reported to applications Depending on your requirements andthe rate of usage, your policy might dictate that you keep disks at less than 50 per-cent or 60 percent capacity at any given time.

You have several options in managing disk capacity, and how and what you choose

to do will vary from server to server, application to application, and business tobusiness The following list highlights the options you have, made possible withWindows 2000 storage services and technology:

✦ Volumes can be spanned, or extended This is the process of chaining volumes

of unallocated space together When a disk approaches the capacity thresholdthat you have set, a new volume can be “bolted on” to extend the capacity Forfaster I/O on spanned volumes, you can also add striping support

✦ Users can be given enforceable disk space quotas, as groups (all users havethe same quota) and as individuals As you will later see, it is possible toassign an across-the-board quota giving all users on a volume the same quota,and thereby create a ceiling no user can violate

✦ You can mount into the file system name space on a given folder This topic isintroduced in this chapter and given full coverage in Chapter 21

✦ You can redirect folders to volumes and root shares on other servers This is

a feature of the so-called distributed files system (Dfs), which is fully covered

in Chapter 21

✦ You can compress data on volumes Compression is covered in Chapter 21

✦ You can create a Hierarchical Storage Management system (HSM) WithWindows 2000 Remote Storage Services (RSS) and Removable StorageServices (RSS), you can automatically move files that are not needed on alocal or network volume to removable media, such as tapes and portabledisks The HSM system is introduced in this chapter and dealt with at length

Check your application’s documentation or call the vendor’s technical support forthe recommended storage requirements.

Figure 16-1: Strengths of the Windows 2000 storage options

Access to Data

The second tier of storage management is maintaining availability If your businessand applications call for 99.9 percent availability — applications and users needuninterrupted access to data — you should consider all available hardware andsoftware options The following services or features of Windows 2000 storage can

be employed to meet the 99.9 percent initiative:

✦ Mirrored volumes: Windows 2000 dynamic volumes allow you to create

RAID-1 compliant mirrored volumes

Response Capacity

✦ RAID-5 fault tolerance: RAID-5 or striping with parity is fully supported You

can build, manage, and break a RAID-5 array with three or more disks

✦ Fail-over servers: The above solutions can be cloned onto hot standby servers

without automatic fail-over Automatic fail-over is a feature of the clusteringservice that is supported in Windows 2000 Advanced Server and Windows

2000 Datacenter Server, which are not covered in this book

Windows 2000’s fault-tolerant dynamic volumes are given comprehensive coverage

in this chapter You can also consider the HMS and Dfs solutions described ously as additional support for the 99.9 percent availability initiative

previ-Protecting Data

Data protection initiatives need to respond to several threats: file corruption, theft,natural disasters, virus attacks, and lost data (user-deleted files, fried hard disks).The primary services that support the backup and restoration of data are as follows:

✦ Removable Storage Services: Discussed later in this chapter and in Chapters

17 and 21, Removable Storage Services work with backup technology, media,and robotics to provide a comprehensive data protection media managementsystem

✦ The Backup/Restore Service: This is fully discussed in Chapter 17, and as

part of the HSM system in Chapter 21

✦ Remote Storage Services (RSS): Also discussed in Chapter 21, the RSS is

responsible for moving online data to offline backup technology RSS, theRemovable Storage Services, and Backup/Restore work in concert to affect

an HSM system

Support for Legacy Systems

Almost all Windows 2000 disk fault tolerance and disk extension is possible via theDisk Management service, which allows you to create dynamic volumes on whichyou can configure extended volumes, mirrored volumes, and RAID-5 arrays You canalso easily manage these volumes However, Windows 2000 dynamic disks are notcompatible with older versions of Windows or any other operating system In otherwords, Windows 95, 98, and NT cannot read from or write to dynamic volumesinstalled as local volumes

Although dynamic volumes can only be locally accessed by Windows 2000, thisdoes not mean you cannot read or write to a share or a publishing folder in ActiveDirectory residing on a dynamic volume

Note

At the time of this writing, there was no way to upgrade or service pack a legacy 9x

or NT 4.0 machine to read a dynamic volume A study of the Windows 2000 tecture suggests that the support for dynamic disks is deep in the new operatingsystem and not something that can be tacked onto NT with a service pack

archi-Disk Management Services

Windows 2000 provides many built-in features that can assist in making a soundstorage plan a reality The operating system provides sophisticated disk and vol-ume management in the form of two disk-centric services: the fault-tolerant diskmanager (FTDISK) and the logical disk manager (LDM) FTDISK and its associatedtools, like FTEDIT, are inherited from Windows NT; it is a basic disk utility

The LDM is a new introduction responsible for the management of dynamic, logicalvolumes Dynamic volumes are storage objects, abstracted above disk spanningpartitions and infused with dynamic or virtual storage attributes that unfortunatelyonly Windows 2000 understands The Windows 2000 disk/volume managementarchitecture is illustrated in Figure 16-2

Figure 16-2: Windows 2000 disk/volume management architecture

FTDISK

RAID-5

MirrorsStriped VolumesSpanned Simple

OTHERS

WINDOWS 2000 AND NTFS 5.0

LOGICAL DISK MANAGER

Employing the services of both FTDISK and the LDM, Windows 2000 storage vices can basically be split into two tiers: Basic Storage and Dynamic Storage Diskscan be configured for one of the two systems The differences and employment ofthe two systems is explained later in this chapter; however, one major benefit ofdynamic storage is the support for mounted volumes.

ser-Mounted Volumes

In terms of dynamic disks and dynamic volume management, mounted volumes vide a means of rapidly adding and removing disk space from a system Installing orimporting a dynamic disk and then mounting the volume anywhere in the file systemnamespace, such as into a folder, is an elegant solution to the problem of adding orremoving disks from a system without shutting down the computer and betrayingthe 99.9 percent availability initiative

pro-Mounted volumes can be constructed out of single dynamic volumes or ant arrays of dynamic disks Because a dynamic volume is an object that encapsu-lates disk space, you can graft dynamic volumes together in various ways to extendand protect the storage space (spanning, mirrors, and so on) Then you can takethe entire extent (capacity) of the volume and graft it into the file system names-pace For example, you can mount the volume onto the d:\disk as d:\dataandthen extend the d:\datavolume as applications, service level, and availabilityrequire

fault-toler-Volumes, and thus the volume mount points, can be extended as needed The onlylimitation is the physical hardware architecture (usually SCSI) and the physicalstorage space for the hard disk

Mounted volumes consist of dynamic disks created in the Disk Manager, discussedlater in this chapter However, you need to become familiar with the advanced fea-tures of NTFS before you can mount volumes into the file system We cover thisimportant topic, volume mount points, in Chapter 21

Disk Defragmentation

Another topic worthy of introduction here is disk defragmentation NTFS is in itself

an orderly and efficient file system, and we, and many administrators we know,have never really needed to defragment an NTFS disk However, disk defragmenta-tion has been introduced into Windows 2000 We discuss it later in the storagehousekeeping section of this chapter and in depth in Chapter 21

Removable Storage

Windows 2000 is the first Microsoft product to introduce a holistic removable age management system built into the operating system Removable storage is anymedia that can be taken out of host equipment and relocated off site Removablestorage typically includes tape drives systems, recordable and rewriteable compactdisks and DVD-ROMs, and even removable hard-disk technology such as ZIP disks,portable or external hard drives, and more

stor-Removable Storage Services allows you to install and configure single tape driveunits (like Quarter Inch Cartridge) or highly sophisticated and expensive roboticlibrary and archive systems (such as CD-R changes and DLT systems) You can alsogather your removable media resources into collections of removable storage,

known as a media pool.

Removable media, such as QIC, DAT, and DLT, are discussed in Chapter 17

Media pools refer to two concepts in backup administration First, they refer to lections of media, regardless of the type, from which a backup system can checkout and check in media required for any instance of backup or restore processing

col-Then there are media pools, a la Windows 2000, which are collections of tapes,carts, and disks organized under a single management system

Media pools are not new to IT, and a stronger definition might be “a collection of likeand unlike storage media collected together as a single storage resource.” WindowsBackup and Remote Storage Services draw on media pools according to the rulesand regimens set up to protect data, and they support 99.9 percent availability

Remote Storage and HSM

Remote storage is the service that makes Windows 2000 look a lot like UNIX HSM isthe widely known (in UNIX and mid-range circles anyway) acronym for HierarchicalStorage Management, which is a major feature of Remote Storage Services, or RSS

Like removable storage services, Windows 2000 is the first Microsoft operating tem to introduce HSM capability

sys-But please do not go running off to your server to dig around in RSS to look for theHSM option, because you will not find it HSM is really a system (some call it a phi-losophy) and consists of numerous components and services that unite to form acascade of continually migrating and translocating data throughout the enterprise

information network (technically speaking, RSS is an HSM system) HSM supports

the 99.9 percent availability mission by moving unused data off local storage when

Note

the disks start getting full, or at pre-configured intervals It can also participate in ageneration-rotation backup system, as discussed in the next chapter, but this is notyet a feature of RSS.

The hierarchical system of data storage has several levels There is a data retrieval

level, which starts out at the top of the hierarchy Data at the top of the hierarchy is

online and always in a highly available or retrievable storage state Data at the tom end of the hierarchy is usually offline and in a latent state of retrieval

bot-High-availability data resides on local fast hard disks, in fast access network storagesilos or storage area networks (SANs), or arrays of hard disks servicing clusters Atthe lower end of the hierarchy, data is stored on tape cartridges, compact disclibraries, tape libraries, and even slower hard disk arrays

How does the data get from the high-availability state to the other end of the chy? This is what you set up in RSS A file that passes a test — such as time sincelast accessed — is moved off the fast front-line volumes and down the hierarchy tothe slower media In a well-designed system, the files that are not being accessedeventually end up on tapes checked into a library The actual file migration can betriggered according to events, such as hard disk space checking, and so on

hierar-The RSS does not move all evidence of the file off the hard disk at the top end, but

leaves an empty file name on the volume as a marker All the data is drawn out of

the file and sent to remote storage From the user’s perspective, the file still exists

on the hard disk When the user needs to access it again in the future, the file ischecked out of the archive system and returned to its original location In fact, theuser need never know the data was actually somewhere else in the system Theonly giveaway would be the slower access time when trying to open a large file that

is in remote storage

The storage level breaks out into two hardware-specific levels The lower level is

the remote storage level, which is responsible for the storage of data that has beendrawn from local storage where it is no longer being accessed Remote storagemonitors file space on the local volume it is managing When available space drops

to a certain level, remote storage will move “stale” files to free up hard disk space.The capabilities of the remote storage system are the following:

✦ You can set up and configure remote storage devices such as tape drives andlibraries

✦ You can configure volumes for RSS management and HSM

✦ You have more disaster recovery options and more tools to protect the 99.9percent availability initiative

✦ You can manage volumes for remote storage and HSM

When remote storage needs to check data into a library, it interfaces with theremovable storage system — via media pools — described earlier to copy the data

to online libraries The combination of the two services (remote and removablestorage) provides the following benefits:

✦ You can deploy low-cost local, remote, and removable storage as a means ofcontinually drawing unused or stale data off the local storage, which needsspace for applications and data This means that as long as you have remov-able storage media, such as carts or compact discs, you can continuously free

up space on the local volumes

✦ You can implement a data backup and archival system that more intelligentlystores data in an archival state (where the data is available at any time) thantraditional backup software You can deploy a systematic data access,archival, and backup plan Backup practice alone is not a perfected science,

as you will see in Chapter 17 The data backup process (speaking generallyand not specifically about the Backup program) leaves a lot to be desired

✦ The HSM system allows you to view all storage resources on your network as

a contiguous storage space like one huge hard drive, hence the term

stor-age area network or SAN Now, your network is not only your computer, it is

also your hard disks as well

Although HSM is not a substitute for traditional regular backup, because there isonly one instance of the data, you might consider installing a system that makesregular, rotation-managed backups of the offline media

Before we get into setting up Remote Storage and an HSM system, we need to becomefully competent with local storage services and Disk Management, Removable Storageand backup, and the NTFS We devote this chapter to local storage, Disk Management,dynamic volumes, and fault tolerance on dynamic volumes You should then becomefamiliar with media pool configuration, backup strategy, and NTFS 5.0, discussed inChapters 17 and 21

The Disk Management Snap-in

This is the perfect time to introduce the Disk Management Snap-in It is loaded intothe Computer Management console, which is located in the Control Panel, underthe Administrative Tools folder You can also load it by executing compmgmt.msc

at the command prompt or by selecting Start ➪ Run and entering compmgmt.msc

The Disk Management Snap-in is a leaf or node on the Storage branch of ComputerManagement You can configure the view of devices from the View menu The snap-in is illustrated in Figure 16-3

Figure 16-3: The Disk Management Snap-in in

Computer Management

The Disk Management Snap-in is the only application you need to manage disks andvolumes in a Windows 2000 system intuitively From it, you can configure basicdisks and partitions, create simple volumes, create and work with spanned vol-umes, create and work with stripe sets, add disks, change storage types, configureremovable storage, and manage media disks on any Windows 2000 remote server.The only thing you cannot do is create fault-tolerant or spanned volumes on basicdisks You can only manage such disks if they were created on NT 4.0 Disks canalso be managed from the command prompt as discussed in Appendix A and withthe services of command line tools like FTEDIT, discussed in Appendix B

You need the appropriate permissions to work with any disk on any computer thatcan be loaded into the snap-in viewer

Figure 16-3 illustrates the Disk Management Snap-in attached to remote computerMCDC02.MCITY.ORG This server started off with one basic local disk (C:) Whilewriting this chapter, we pulled two old Quantum Vikings with 4.5MB SCSI disk drivesout of a storage (shoe box storage) and installed them into MCDC02 The pair ofdisks was impossible to install on Windows NT The computer would recognize theAdaptec controller card, but always tossed us a Blue Screen of Death (BSOD) uponboot up It was something about the hardware NT did not like, and the Adaptec-suggested workaround was more expensive than buying two new disks

On Windows 2000, however, the drives installed with no pain at all The operatingsystem immediately detected the foreign drives and installed them into the system

as if they were Plug and Play At first, Windows 2000 recognized the drives as basicpartitions The disk list and volume list views are illustrated in Figure 16-3

Note

Before we proceed, you should understand that you cannot set up a disk to be both

a basic and a dynamic disk, but you can set up a Windows 2000 server to supportand write to both basic and dynamic disks in the same machine This is important ifyou plan to continue running legacy Windows servers and clients, because theycannot directly access the new dynamic volumes

Basic Storage

Basic storage is the industry standard Hard disks are divided into a system of titions A partition is essentially a part of a physical disk’s total storage space thatacts as a separate standalone unit When you install a hard disk, you have to firstdivide it into partitions You chop the physical disk space into a primary partitionand assign it a hard drive letter such as C:, and a system of extended partitions can

par-be set up as logical drives (assigned any drive letter other than the letter the mary partition is using)

pri-Disks that are set up for basic storage are known as basic disks Basic disks can

con-sist of primary partitions, extended partitions, and logical drives Basic disks cancontain up to four partitions You can set up the disk to contain four primary parti-tions, or three primary partitions and one extended partition Basic disk architec-ture is illustrated in Figure 16-4

Figure 16-4: Basic disk architecture

DISK

primary partitions partitionsprimary

DISK

extended portions

F G H C

Primary partitions cannot be further divided into logical drives, but extended

parti-tions can Any primary partition can be active, which means it can contain a boot

sector from which operating systems can be started Primary partitions can also betiny, just enough to hold the necessary files to start the OS For example, on earlyNetWare servers, the primary needed to be no larger than 30MB, just enough toboot the server The remaining extended space was then configured as an extensionholding the actual operating system, the NetWare system (SYS) volumes

Basic storage (on Windows NT 4.0 and earlier) provided us with only centric disk management architecture You could (on the server platform) createmirrored sets, stripe sets, spanned sets, and RAID-5 volumes On Windows 2000,under basic management and using FTDISK, these sets can be broken up in thesame fashion as you did on Windows NT, which also includes having to restart theserver after configuration of the sets But as mentioned earlier, these disks cannot

partition-be created by Windows 2000 Disk Management

Dynamic disks cannot and do not contain partitions or logical drives (these areunmounted when you upgrade a disk to dynamic storage) The LDM sees the entireunderlying disk as one whole partition and then encapsulates the storage object ontop of that

Primary Partitions

The active primary partition is the part of a disk on which you start the operatingsystem In other words, the primary active partition is where you install the bootfiles that start the operating system On dual boot or multi-boot systems, you caninstall several operating systems, and different versions of the same operating system, in the active partition

You can only boot the operating system if it knows how to work with the underlyingfile system In other words, you cannot boot to Windows 95 or Windows 98 on anNTFS file system, but you can boot Windows NT or Windows 2000 from a FAT system File systems are discussed in depth in Chapter 21

Extended Partitions

Extended partitions are created out of the free space that exists after all primary partitions have been created There can only be one extended partition, which willinclude all the remaining free space left on a disk after you create a primary partition

Extended partitions are divided into segments (one or more), and each segment can

be set up as a logical or virtual drive You do this from within Windows 2000 UsingWindows 2000 disk management and storage tools, extended partition segments can

be configured to represent logical drives, and they can be assigned drive letters

System Partitions

As described in Chapter 5, the most common setup for Windows 2000 is to create apartition that can store boot files, and then configure the remaining space as a sin-gle extended data partition We usually create a “system” partition and a “data”

partition on Windows NT, but you might not be able to do that if Windows 2000 isallowed carte blanche access to all available storage Windows 2000 refers to thepartition it boots from as the System partition, which is the converse of all other,non-Microsoft operating systems

When you install Windows 2000, it creates a bootable partition (in the system

parti-tion) that contains only the necessary hardware- and OS-specific files that arerequired to start the operating system You can then install all system files intoanother disk’s primary partition or an extended partition This makes for someinteresting recovery possibilities, which we will discuss later in Appendix B

When you first install and boot to Windows 2000, all disks are configured as basic,and the fault tolerance is created and managed by FTDISK However, basic volumes,once configured, remain fixed They can only be converted or upgraded to dynamicdisks, where they become the responsibility of the LDM

The conversion to dynamic storage takes place from within Windows 2000, usingthe Computer Management snap-in

Dynamic Storage

Windows 2000 now supports a standard known as dynamic storage in which a gle disk is configured as a single active partition This configuration is known as a

sin-dynamic disk Dynamic disks are not created from the raw drive, but are created by

converting a basic disk

Dynamic disks support volumes, which are storage objects that can consist of portions of a single dynamic disk or several dynamic disks The cool thing aboutdynamic disks is that they can be configured into several different types of vol-umes, from simple volumes to redundant arrays of volumes supporting various lev-els of fault tolerance and striping The volume types are simple, spanned, mirrored,striped, and RAID-5 level

The LDM brings the following features to Windows 2000 volume management:

✦ Online Management: RAID configuration can be created, broken, and rebuilt

without rebooting the system This feature allows you to install a working tem, meet service level requirements as soon as the system is available, andthen continue to work on the running system to meet data protection and stor-age requirements, while the system is live and servicing users and applications

sys-✦ Extendable File Systems: Volumes can be extended or mounted while the

system remains up

✦ Self-identification: LDM disks are self-identifying In other words, once a disk

is configured, it can be removed from one system and added to another tem or host while both systems are hot Differences in the target system, such

sys-as SCSI addressing, controllers, and so on, do not compromise the disk (seeImporting Volumes, discussed later)

✦ MMC: The LDM’s GUI is the Disk Management Snap-in You can delegate the

disk administrator task and use Disk Manager to attach to the storage vices of remote computers

ser-As mentioned previously, the wonderful thing about dynamic disks is that they can be managed, even resized and added to, without shutting down, rebooting, orrestarting the operating system The RAID-5 level volumes support the full RAID-5standard and are thus ideal for storage arrays and mission critical systems

If your application or server calls for Windows 2000 exclusive use, the servershould be set up with dynamic disk volumes to obtain the full benefits of Windows

2000 storage features and the distributed features of both storage and the file tem The various installation options you have are covered in Chapter 5 For thetime being, Table 16-1 sheds light on the differences, as they affect you, betweenbasic and dynamic disks

sys-Table 16-1

Basic Disks

Windows

Windows

Dynamic Disks

The disks that are managed by the LDM service are called dynamic disks The LDMdoes not really touch the physical disk or the drivers, but rather works with logicaldisk objects that are an abstraction layer above low-level system code that goesdown various layers, via drivers, to the physical disk itself The logical-dynamicdisks are similar in concept to the logical printers in Windows 2000 You do notactually manage the disk, but rather you interface with the object, which acts as aservice proxy, setting and changing various properties You do not do this directly;

it is done by the Disk Management Snap-in discussed earlier

Microsoft has chosen to offer limited support for fault tolerance on basic disks Youneed to upgrade disks to dynamic status to get the full suite of software fault-tolerance functionality It thus appears that Microsoft is being obtuse with respect

to backward compatibility, because Windows NT and Windows 98 and earlier not read dynamic volumes installed into the local host either But we believe thelack of support is more a case of incompatible code Our guess is that Microsoft is

can-at work on an API thcan-at will change thcan-at If you want to get the full functionality of

Windows 2000 storage services, upgrade those NT 4.0 servers and 9x clients.

The dynamic disk objects are managed as groups, or as a collection of disk objects

When you set them up for fault tolerance, each disk manages a tiny database on itsmedia that contains information about all the other disks in the host Thus, anychanges to the disk configurations that control the fault-tolerance mechanics of thedisks, such as mirror and parity information, are updated to all the disks Disk config-uration is thus no longer maintained in the registry, and the configuration databasesare not currently accessible through any formal storage API

The disk groups can be collections of simple volumes, concatenated or extendedvolumes, stripe sets, mirror sets (RAID-1), and stripes sets with parity (RAID-5) Thelogical disk architecture thus allows the volumes to be extended at any time with-out rebooting the host

Note

You can convert basic disks to dynamic disks, and you can convert them back again

to basic disks You do not lose any data converting to a dynamic disk, but you dolose your data if you convert back again You should, therefore, fully understandthe differences and benefits, before you convert, because for the most part, it’s aone-way street It is also worth repeating that Windows NT cannot directly compre-hend dynamic disk volumes

Also, in order to understand how dynamic disk configuration and the five volume

types apply to your storage plans, you should first understand each volume type.

It is also important to understand that although you can promote a basic disk to adynamic disk, any hard partitions created, especially on primary active system andboot partitions, cannot be extended You first have to delete the partitions and con-figure the disk as a basic disk, then promote it But, deleting the partitions will nodoubt destroy the data on the disk, so you need to either copy the data to anotherdisk or back it up to remote or removable storage If you have applications, share-points, and configurations that are particular to the disk you need to convert, youwill have to plan this very carefully

Installing Windows 2000 with more than one disk in the system may be a problem

in certain situations, because you cannot make disks fault-tolerant if a partitionexisted before you created the dynamic disk This is highlighted and a workaround

is provided in Chapter 5

The five dynamic volume types

The following list describes the five volume types, from simple to RAID-5:

✦ Simple volumes: A simple volume is disk space configured from a single disk.

Simple volumes on their own do not provide much benefit, but do provide astarting point to extend storage space on the local system You need to startwith simple volumes if you are aiming to extend volumes, span volumes, orprepare for fault tolerance in the near future Simple volumes can also bemounted to NTFS folders to increase local storage on the server without shut-ting down the server or affecting service level and users (Mounted volumesare discussed in detail in Chapter 21.)

✦ Spanned volumes: Take simple volumes and chain them together to extend

the hard disk space on a local volume, and you have a spanned volume set.Spanned volumes can consist of between 2 and 32 physical disks in the samesystem Spanning is supported by NTFS and the FAT32 file system

Windows 2000 writes the first disk and when that disk is full, it simply writes

to the second disk and so on until it runs out of space on the 32nd disk.Spanned volumes are not fault-tolerant, so if any disk in the span breaks, thedata on that disk is lost

Tip

✦ Striped volumes: These volumes provide similar service to spanned volumes,

but all the disks are configured as a single volume, tantamount to having onedrive C: consisting of 32 disks Striped volumes are used more for efficiencyand for extending storage The operating system writes to any free space onany of the disks in the set Disk I/O is more efficient

If one disk in the volume breaks, you lose the data in the entire volume Stripedvolume sets are essentially RAID level 0 (Redundant Array of Inexpensive Disks)

or RAID-0, in that you create an array of disks but you do not provide for any

fault tolerance

Striped volumes seem set to become obsolete because the concept was born

in the days when hard disks were more expensive and did not have the ity and bandwidth they have today, and will have in the future Today’s disksalso read and write data much faster than anyone would have believed possi-ble in the last decade (SCSI-2 drive platters, for example, have already reached10,000 RPM) Also, the risk of having one disk failure take out the entire array

capac-is far too ghastly an idea to contemplate If you use the full array set (32 10GBdisks), one disk failure could potentially cause irrecoverable harm Can yousleep well at night knowing that any one of 32 drives in some absurdly hugearray might fail at any time?

✦ RAID-1 volumes (mirroring): This is the first level of software fault tolerance

provided by Windows 2000 Essentially, a mirrored volume set consists of twoidentical copies of a simple volume If one hard disk fails, you can carry onworking on the survivor until a new simple volume is added back to the mir-ror Mirrored volumes are ideal for providing inexpensive fail-over potential,

as described in Chapter 17 on disaster recovery and backup/restore youcould call a mirrored volume set “the poor admin’s fail-over.”

✦ RAID-5 volumes (fault-tolerant arrays): These volumes are Windows 2000

software fault-tolerant striped redundant array (RAID) volume sets The ating system adds a parity-information stripe to each disk partition in the vol-ume This stripe is used to reconstruct the data when a disk in the arraybreaks You need at least three disks of any size to construct a RAID-5 volume

oper-RAID-5 volume sets provide two benefits; they stripe data across the entirelogical drive consisting of all the disks, and the data on any lost disk is recov-erable because it is rebuilt from data on the remaining drives

The fault-tolerant volume sets deserve more coverage

RAID-1: Disk Mirroring

Disk mirroring is the process of creating a 100-percent identical “clone” of a harddisk, hence the term “mirror.” The mirroring process is done in real time or concur-rently to both disks When you create a mirrored set, both disks are configuredunder the same drive letter and appear as a single disk to the OS, so when one diskgoes belly up, the other can continue regardless