Fault tolerance and disaster recovery Unit objectives Discuss disk configuration Discuss...

54
Fault tolerance and disaster recovery Unit objectives • Discuss disk configuration • Discuss Windows-based replication and NDS/eDirectory partitions and replicas • Discuss backup and UPS

Transcript of Fault tolerance and disaster recovery Unit objectives Discuss disk configuration Discuss...

Fault tolerance and disaster recovery

Unit objectives

• Discuss disk configuration

• Discuss Windows-based replication and NDS/eDirectory partitions and replicas

• Discuss backup and UPS

Topic A• Topic A: System fault tolerance

• Topic B: Replication

• Topic C: Backup and UPS

Disaster planning• When creating a disaster plan, some key

points to be considered are:– Plan for the worst– Implement physical data security– Protect your critical systems

RAID• A set of specifications describing hard disk fault

tolerance configurations• The specification:

– RAID Level 0– RAID Level 1– RAID Level 2– RAID Level 3– RAID Level 4– RAID Level 5

• There is also a RAID 10 (as in RAID “one plus zero”), a RAID 01, and several other “composite” levels.

RAID• Both the book and the power point are weak on

RAID; the power point also presents the material out of order – so I’ve included 10 slides on RAID from other power points, with editing:

• Most popular levels are RAID 0, 1, 5 and 10.• RAID = “Redundant Array of Inexpensive or

Independent Disks.”• More disks give more heads give faster

transfer rates.• Sometimes the “read” is faster and sometimes

the “write” is faster; sometimes both are faster.

RAID 0• Disk striping (no parity): (data is written across disks in a stripe).

• Stripe size is a multiple of 2, size depends on the RAID level in use, RAID 1 and 0 have a high stripe size, usually 128K whereas RAID 5 has a low stripe size usually 16K.

• RAID 0 offers no redundancy (no fault tolerance).

• The 2 128K stripes of data are written in parallel.

• Note that because you have the same amount of data, but written across 2 or more disks, you have more locations from which to read it. This results in faster disk reads – you have 2 heads reading at once.

• “Disk writes” are also faster; you have 2 places to which to write!

• Disadvantage: There is no fault tolerance, because there is no parity.

Disk 0

Disk 1

Controller

256K Data

128K Data

128K Data

RAID 1 (Mirroring)• Minimum of 2 disks, literally one disk is a

complete mirror of the other.• One fails the other takes over• When reading can read both disks (two copies of

the data) – very fast read and write access.• Um, no – reads are faster – you have 2 identical

disks from which to read, but you have to write everything twice! This isn’t as slow as it seems, especially when using 2 controllers – duplexing – but it isn’t fast either.

Disk 0

Disk 1

Controller

256K Data

256K Data

256K Data

RAID 1 (Duplexing)• Controller Mirroring (2 controllers) each with a disk• Just like mirroring but with two controllers instead

of one, so if one controller goes down, you still have one disk that is fine.

Software mirroring (RAID1)

Disk 0

Disk 1

Controller

Disk 0

Disk 1

Controller

RAID 10 or RAID 01

• RAID 10 is mirroring (1) then striping (0)

• RAID 01 is striping (0) then mirroring (1)

Disk 0

Disk 1

256K Data

Disk 2

Disk 3

128K Data

RAID 1+0

128K Data

1. Write Single Stripe2. Mirror3. Write Single Stripe4. Mirror

Disk 0

Disk 1

Disk 2

Disk 3

128K Data

RAID 0+1

128K Data

1. Write Single Stripe2. Write Single Stripe3. Mirror

256K Data

RAID 10 or RAID 01• Minimum of 4 disks because data needs mirroring

and striping.• Massive difference when comes to fault tolerance

so be careful!• RAID 10 allows for more fault tolerance – any disk

can go so long as it has a mirror.• RAID 01 has poor fault tolerance – loose 1 disk in

both mirrors and array fails.• Make absolutely sure you are getting what you think

you are getting – there is a difference!!

RAID 5• Minimum of 3 disks required.• Uses parity to recalculate data in case of disk failure

– an EOR formula = “exclusive or” 1+0 or 0+1 = 1, 0+0 or 1+1 = 0

• Critical failure occurs on failure of 2 disks.• Performance degradation on single disk failure.• Uses a smaller stripe size to aid parity calculation.

Disk 0

Disk 1

Controller

256K Data

16K Data

16K Data

Disk 116K Parity

RAID 5 Data Calculation• EOR to calc parity and EOR to re-calc data

Stripe 1 goes to Disk 1 -> 10101010 (170)Stripe 2 goes to Disk 2 -> 10111101 (189)Parity Stripe to Disk 3 is 10101010 EOR

10111101------------00010111 (23)

00010111 is written to disk 3 .

Recovery (Disk 2 has failed)…Take data from Disk 1 10101010 EORTake parity from Disk 3 00010111

------------Data on Disk 2 is 10111101

Disk mirroring (back to Thomson book)

Activity A-1 – page 20-5

Discussing RAID

Disk duplexing

Activity A-2 – page 20-7

Discussing disk duplexing

Disk striping with parity• An implementation of RAID Level 5

• Normally used on larger networks where data integrity is a critical concern

Activity A-3 – page 20-9

Discussing disk striping with parity

Volume sets

• Combines space from up to 32 drives• Cannot contain the system or the boot partition• If one disk area is destroyed, the entire set fails• This is the simpler version of a “striped set”

Disk striping

• Also combines space from up to 32 drives

• Each segment must be the same size

Activity A-4 – page 20-11

Managing disk configuration

Topic B• Topic A: System fault tolerance

• Topic B: Replication

• Topic C: Backup and UPS

Replication• Offers additional data redundancy on

Windows-based networks

• Can specify certain data to be copied from one system to another

• Common uses include – replication of login scripts to all domain servers– replication of mandatory user profiles– replication of frequently used files across multiples

servers to balance the server load

Replication• Available in Windows NT networks

• Helps to copy data automatically from a source system (exporter) to a destination system (importer)

Key points about replication• Runs as a background service• After any changes, files must be closed before they

can be replicated• Can specify to replicate files immediately after a

change in the subdirectory tree• Individual subdirectories might be locked• An exporter can send files to importers• An importer can receive files from one or more

exporters• The import directory might be locked• A Windows NT Server might act as both an exporter

and an importer

Active Directory• Fault tolerance of directory services information

is built into the directory model

• Every domain controller holds a copy of Active Directory

• So, by this syllogism, fault tolerance is “assured”– All domain controllers contain Active Directory– All Active Directory provides fault tolerance– Therefore, all domain controllers provide fault

tolerance–

File Replication Service• In Windows 2000/Server 2003, the File

Replication Service (FRS) replaces the LAN Manager Replication system used in Windows NT

• Used to replicate system policies as well as login scripts.

• Allows for file replication for domain-based Distributed File System (DFS).

Activity B-1 - page 20-13

Discussing replication

NDS / eDirectory partitions• Involves division of NDS/e-Directory database

• Provides two primary benefits:– Fault tolerance– Performance Increase

NDS/e-Directory Partitions & Replicas• Used to store information about all of the

objects known to the network• A partition is a logical division of the eDirectory

database. A directory partition forms a distinct unit of data in the tree that stores directory information.

• Partitions can be created at container level objects, like Organization, Organizational Unit or any objects marked as a container.

• An eDirectory has one [ROOT] partition which contains all the objects by default.

• Partitions are set up as parent-child objects.

NDS/eDirectory directory partitions

Activity B-2

Managing NDS/eDirectory partitions

NDS/eDirectory replicas• A replica is a copy or an instance of a user-

defined partition that is distributed to a server

• Each partition has at least one replica

• Examples of Types:– Master replica– Read/write replica– Read-only replica– Subordinate reference

NDS/eDirectory Replica Types• There are six types of replicas:• 1. Master replica: There can be only one Master

replica for a partition. The Master is a read-writeable replica that, most importantly, controls the partition operations and the obituary process.

• This type of replica also performs the following operations:– Managing objects (add, remove, move)– Authenticating objects– Managing attributes (add, remove)– By default the first server in the tree holds the Master

replica of the [ROOT] partition.

NDS/eDirectory Replica Types• 2. Read-Write replica: This replica type allows

modification to objects and will automatically propagate them to the other replicas based on the timestamps. – You can designate a Read-Write replica as a

Master replica.

• 3. Read-Only replica: This replica type is only readable. – It does not perform any write operations– It will forward all writing requests to a Read-Write

replica. – The replica can be designated as a Master replica.

NDS/eDirectory Replica Types• 4. Filtered Read-Write Replica: This replica

contains only a special set of classes and attributes specified by the filter.

• The replica can be written and the changes will be synchronized to the other replicas.

• 5. Filtered Read-Only Replica: The same rules applies to this replica type as the ones to the Filtered Read-Write Replica, but the replica is only readable

• Therefore all writing requests are forwarded to a writeable replica.

NDS/eDirectory Replica Types• 6. Subordinate reference replica: System-generated

replicas that don't contain all the objects, attributes and values like a master or a read/write replica.

• Therefore, they don't provide fault tolerance. • They are internal pointers generated to contain enough

information for eDirectory to resolve object names across partition boundaries.

• You cannot create a Subordinate references replica; eDirectory will create it when the server holds a replica of the parent partition, but not one of the child partitions.

• It holds no partition data, only information about the "real" replica-holder servers.

• So it cannot be designated as a Master without adding a Read-Write or Read-Only replica.

Activity B-3 - Page 20-17

Discussing NDS/eDirectory replicas

Replica ring• Made up by the servers that hold replicas for

that partition

• Documentation of the replica ring might consist of a replica table containing:– A list of servers– A list of partitions– The type of replica stored on each server

Activity B-4

Discussing replica rings

Fault tolerance• Topic A: System fault tolerance

• Topic B: Replication

• Topic C: Backup and UPS

Backup and UPS• Backup of the data is necessary to protect the

data in the event of loss of power

• Three types of backup schemes:– Full backup– Incremental backup– Differential backup

Activity C-1

Discussing backup schemes

Removable media• An alternative to tapes

• Includes:– Removable hard drives– Floptical media– Rewriteable CD-ROMs and DVDs

• Provide a convenient way to archive data

Activity C-2

Discussing removable media

Backup storage• Storing backups in your office isn’t a good idea

• Always keep backups in a secure, access-controlled location

• Also have backups stored at offsite locations

Disaster recovery site options• Cold sites

• Warm sites

• Hot sites

Cold sites• Cold site

– Usually a single room in which your data center can be recreated in case of a disaster

– Can be on site or off site– Doesn’t actually hold any equipment– Coming back on line after a disaster can take quite

a bit of time– Least expensive backup site solution

Warm sites• Warm site

– Can be either on site or off site– Contains a fair amount of equipment to create a

semi-duplicate of your current data center – Can be live in much less time than a cold site– Is more expensive to create and maintain than a

cold site

Hot sites• Hot site

– Is a complete duplication of your current data center– Is typically off site – Can be up and running in a matter of hours– Very expensive to create and maintain

Activity C-3

Discussing backup storage

Uninterruptible power supply• Makes sure that the server is powered down,

thereby protecting network data

• Keeps your system from going down unexpectedly due to line power loss

• Uses a battery to supply power to the system

Activity C-4

Discussing UPS

Unit summary• Discussed disk configuration

• Discussed Windows replication and NDS/eDirectory partitions and replication

• Discussed backup and UPS