Rapid Backup andRecovery Strategies

for NT/Linux

Matthew F. ReaganMatthew F. Reagan

Centocor, Inc.Centocor, Inc.

mreagan@fast.netmreagan@fast.net

PurposePurpose

The purpose of this presentation is The purpose of this presentation is to identify a problem with to identify a problem with traditional backup and recovery traditional backup and recovery strategies and propose alternative strategies and propose alternative or supplemental solutionsor supplemental solutions

Presentation GoalsPresentation Goals

Discuss the problems encounteredDiscuss the problems encountered Identify alternative or supplemental Identify alternative or supplemental

solutionssolutions Propose a supplemental solutionPropose a supplemental solution Present a snapshot of research into Present a snapshot of research into

rapid backup and recovery solutionsrapid backup and recovery solutions Emphasis on non-production Emphasis on non-production

environmentsenvironments

Presentation Presentation NonNon-Goals-Goals

Present the Present the onlyonly solution solution Present the Present the bestbest solution solution Present the Present the cheapestcheapest solution solution Present the Present the fastestfastest solution solution Ignore the subjective side of the Ignore the subjective side of the

issueissue Replace backupsReplace backups

Production EnvironmentsProduction Environments

Standard hardware configurationsStandard hardware configurations OS installed using SOPsOS installed using SOPs Documented software installationDocumented software installation Usually static configurationsUsually static configurations Centralized, regular backupCentralized, regular backup

Non-production Non-production EnvironmentsEnvironments

DevelopmentDevelopment ValidationValidation EvaluationEvaluation SandboxSandbox

Non-production Non-production Environments (cont)Environments (cont)

Not always standard hardwareNot always standard hardware Baseline OS/software configurationBaseline OS/software configuration Myriad extensions to the baselineMyriad extensions to the baseline Loose documentation Loose documentation

requirementsrequirements Usually not backed up (regularly)Usually not backed up (regularly)

Description of ProblemDescription of Problem

How do you recover from media How do you recover from media failurefailure

How do you recover from accidentsHow do you recover from accidents How do you recover from bad How do you recover from bad

installsinstalls … … with YOUR deadlines?with YOUR deadlines?

Checklist for System Checklist for System RecoveryRecovery

Repair the hardwareRepair the hardware Reinstall NT ServerReinstall NT Server Reinstall the Service Pack(s) as Reinstall the Service Pack(s) as

NeededNeeded Reinstall the required driversReinstall the required drivers Reinstall the appropriate applicationsReinstall the appropriate applications Recover the OS from backup tapeRecover the OS from backup tape

Problems with RecoveryProblems with Recovery

Requires installation and Requires installation and configuration knowledge ofconfiguration knowledge of– NT ServerNT Server– Service PacksService Packs– OS baselineOS baseline– Application SoftwareApplication Software– Drivers (you know where they are, Drivers (you know where they are,

right?)right?)

Rapid Recovery OptionsRapid Recovery Options

Implement robust backupsImplement robust backups RAIDRAID Drive “snapshots”Drive “snapshots”

Implement robust backupsImplement robust backups

Requires additional hardware (tape Requires additional hardware (tape drives)drives)

Requires additional software Requires additional software (backup)(backup)

Requires (more) operations staffRequires (more) operations staff Train personnel in NT, service pack, Train personnel in NT, service pack,

driver and software installation and driver and software installation and configurationconfiguration

Implement robust backups Implement robust backups (cont)(cont)

Recovery requires extensive Recovery requires extensive configuration and installation configuration and installation knowledgeknowledge

Recovery is complexRecovery is complex Recovery is time consumingRecovery is time consuming

Implement local backupsImplement local backups

Requires additional hardware (tape Requires additional hardware (tape drives) on each serverdrives) on each server

Requires additional software (backup)Requires additional software (backup) Train personnel in backup proceduresTrain personnel in backup procedures Needs tape storage/retention policiesNeeds tape storage/retention policies (not much better than previous option)(not much better than previous option)

RAIDRAID

Requires additional hardware Requires additional hardware (drives)(drives)

Requires additional software Requires additional software (backup)(backup)

Train personnel in RAID Train personnel in RAID installation, configuration and installation, configuration and operationoperation

RAID (cont)RAID (cont)

Recovery requires extensive Recovery requires extensive configuration and installation configuration and installation knowledgeknowledge

Recovery is complexRecovery is complex Recovery is time consumingRecovery is time consuming Implementation nightmareImplementation nightmare

Drive “Snapshots”Drive “Snapshots”

Wouldn’t it be nice if we could just Wouldn’t it be nice if we could just take a snapshot of the hard drive?take a snapshot of the hard drive?

Imagine, if you will,Imagine, if you will,taking a snapshot...taking a snapshot...

Prior to installing a service packPrior to installing a service pack Prior to installing a new software Prior to installing a new software

versionversion Prior to modifying an existing Prior to modifying an existing

configurationconfiguration Just after finishing an installJust after finishing an install

Recovery would be…Recovery would be…

Simply putting back the snapshotSimply putting back the snapshot

Functional RequirementsFunctional Requirements

OS independentOS independent Floppy basedFloppy based

– Avoids open filesAvoids open files– Avoids configuration differencesAvoids configuration differences

Simple to useSimple to use CheapCheap

PowerQuest Drive ImagePowerQuest Drive Image

Provides a drive snapshot capabilityProvides a drive snapshot capability Supports DOS, Windows 3.x, Windows Supports DOS, Windows 3.x, Windows

95, Windows NT, Linux95, Windows NT, Linux Reads IDE and SCSI hard drivesReads IDE and SCSI hard drives Writes to IDE and SCSI hard drives, Writes to IDE and SCSI hard drives,

ZIP, Jaz drives, network drives, etc.ZIP, Jaz drives, network drives, etc. Creates an optionally compressed Creates an optionally compressed

image file (~40%)image file (~40%)

Possible ConfigurationsPossible Configurations

Local hard driveLocal hard drive Internal/External ZIP driveInternal/External ZIP drive Internal/External JAZ driveInternal/External JAZ drive Network driveNetwork drive

Local Hard DriveLocal Hard Drive

Destination drive must be Destination drive must be mountable under boot floppy mountable under boot floppy (native IDE, ASPI driver for SCSI)(native IDE, ASPI driver for SCSI)

Destination drive must have a FAT Destination drive must have a FAT partitionpartition

Boot using supplied floppyBoot using supplied floppy Point and click snapshot done!Point and click snapshot done!

Internal/External ZIP DriveInternal/External ZIP Drive

Internal uses IDE or SCSIInternal uses IDE or SCSI External uses PP or SCSIExternal uses PP or SCSI Destination cartridge must have a Destination cartridge must have a

FAT partitionFAT partition Boot using supplied floppyBoot using supplied floppy Point and click snapshot done!Point and click snapshot done!

Internal/External JAZ DriveInternal/External JAZ Drive

Internal uses SCSIInternal uses SCSI External uses PP (traveler) or SCSIExternal uses PP (traveler) or SCSI Destination cartridge must have a Destination cartridge must have a

FAT partitionFAT partition Boot using supplied floppyBoot using supplied floppy Point and click snapshot done!Point and click snapshot done!

Network DriveNetwork Drive

Create a network boot floppy (see Create a network boot floppy (see NT client administrator)NT client administrator)

Must have write privs on shareMust have write privs on share Boot using network boot floppyBoot using network boot floppy Point and click snapshot done!Point and click snapshot done! (took 5 minutes for 2 gb partition (took 5 minutes for 2 gb partition

with 50% of disk in usewith 50% of disk in use

RecoveryRecovery

Boot using supplied (or network) Boot using supplied (or network) floppyfloppy

Gain access to the imageGain access to the image Run PQDIRun PQDI Point and click snapshot recovered!Point and click snapshot recovered! Network recovery took 5 minutes for Network recovery took 5 minutes for

a 2 gb partition with 50% of space in a 2 gb partition with 50% of space in useuse

CaveatsCaveats

Only supports non-RAID Only supports non-RAID configurationsconfigurations

Only supports Intel (or clone) CPUsOnly supports Intel (or clone) CPUs Have not tested larger than 6.4 gb Have not tested larger than 6.4 gb

HDsHDs There are other options There are other options

(Symantec’s Ghost)(Symantec’s Ghost)

ConclusionsConclusions

Several untraditional options Several untraditional options availableavailable

Each is a balance between cost, Each is a balance between cost, convenience, and efficacyconvenience, and efficacy

Drive Image is a strong contenderDrive Image is a strong contender Additional research is necessaryAdditional research is necessary

Questions?Questions?