65
PowerHA for IBM i Solutions Portfolio Allyn Walsh Consulting IT Specialist •Power Systems Strategic Initiatives [email protected]

PowerHA SystemMirror for IBM i - Gateway/400...IBM i 7.1 / 7.2 PowerHA SystemMirror for i Metro Mirror Synch DS8000 • SVC • V7000 V3700 • V5000 • V840/v9000 Flash Copy •DS8000

  • Upload
    others

  • View
    12

  • Download
    1

Embed Size (px)

Citation preview

PowerHA for IBM i Solutions PortfolioAllyn WalshConsulting IT Specialist•Power Systems Strategic Initiatives [email protected]

1

© Copyright IBM Corporation 2016.

PowerHA SystemMirror

Power Systems High Availability Solution For mission critical application availability through planned and unplanned outage events

Shared Storage Clustering Technology designed for automation and minimal IT operations.

Multi-Site Clustering for enterprise class availability

Embedded for integrated simplicity and reliability

2

© Copyright IBM Corporation 2016.

PowerHA SystemMirror Complete IBM Power Systems integrated end to end solutions for HA DR

• Focus: 24 x 7 Application availability through planned or unplanned outage events• Deeply integrated extension of IBM i (implemented in LIC and the OS)

Clustering technology • Provides the applications with a complete resiliency infrastructure• Monitors and manages primary and secondary resources for HA and DR operations

Storage based data resiliency• Data resiliency is an extension of the host system storage management architecture• Storage volumes are either switchable or mirrored between nodes in the cluster• Hardware based replication services for Multi-Site Operations

• Host Based Replication (Geographic mirroring for IBM i, or GLVM for AIX)• Storage Base Replication (Metro Mirror or Global Mirror)

Overall solution characteristic • Automation, minimal IT operations involvement• Data between primary and secondary nodes always in sync always ready for a failover event

3

© Copyright IBM Corporation 2016.

PowerHA Business Value & Objectives

• Provide IBM i Clients With Confidence & Certainly of Outcome • PowerHA eliminates the classic logical replication dilemmas:

• Have we identified all of the objects that need to be replicated ?• Are all of those objects in sync ? Should we or should we not failover ?

• ROI Acquisition and Cost of Ownership• PowerHA is priced per processor core / CBU reduces total number of licenses required• FlashCopy (save window elimination)• Minimize operational expense through automation

• Enable best of breed performance and HA/DR metrics• Provide HA/DR and application performance• Exploit and integrate modern storage technology• Eliminate single points of failure• Enable RPO of 0 and best case RTO

• Native Embedded Technology for HA/DR Solutions• PowerHA technology is an integrated extension of SLIC and IBM i

4

© Copyright IBM Corporation 2016.

Causes for Production Downtime

• Software backups

• Firmware, hardware maintenance

• Software maintenance / upgrades

• User error

• Hardware / software errors

• Disasters• Man-made• Natural

5

© Copyright IBM Corporation 2016.

IBM i HA/DR Customer Profile

• Internal Disk• Logical replication• Role-swap = seldom if ever• Backup window = ranges up to a few

hours• TCO = staffing, SWMA, upgrade

charges, • Staffing = dedicated specialist• Outage management = complex• Support = vendor

• IBM Storage or (Internal Disk with Geomirroring)

• Switchable LUNs, Metro or Global Mirroring with IBM SAN

• Role-swap = daily, weekly, monthly, quarterly

• Backup window = minutes (Flash Copy)

• TCO = minimal staffing, SWMA • Staffing = minutes/week • Outage management = simple • Support = IBM

Legacy solution PowerHA

6

© Copyright IBM Corporation 2016.

Internal vs External Storage - considerations

• What needs to be considered to determine which type of disk storage is "right" for your IBM i environment?

• Size of data in the enterprise – isolate or consolidate?• I/O performance• Availability and Backup requirements• Frequency of deploying new partitions or workloads• Appetite for flexibility vs. complexity• Cost

• Possible advantages of external storage• Flashcopy to offload backups• Elimination of hardware maintenance downtime with Live Partition Mobility• Other virtualization technologies such as Active Memory Sharing, Suspend/Resume

• Recent announcements simplifying external storage deployment• Storwize models – economical but still full function• Flash Storage – V9000 or V840, or Flash systems behind a SVC or Storwize• Native attach options

7

© Copyright IBM Corporation 2016.

For PowerHA - Who’s Doing the Replication?

IBM i• Technologies: Synchronous and Asynchronous Geographic Mirroring• Storage agnostic, although predominantly used with internal storage• Source and target could be different storage types

• Ie. one side could be SAN and the other side Internal disk. • Would allow exploiting SAN capabilities (ie. FlashCopy)

• System failure affects production workload and data replication

External Storage Server• Technologies: Metro Mirror, Global Mirror, LUN switching, FlashCopy• PowerHA supports specific external storage devices

• DS8K, San Volume Controller (SVC), Storwize storage servers, V840 and V9000• Data replication independent of IBM i• Replication overhead offloaded to external storage device• Additional external storage functionality available, such as FlashCopy

8

© Copyright IBM Corporation 2016.

PowerHA SystemMirror IASP Replication Technologies

1 site Shared Storage

External Storage

PROD HA

Network

External Storage

Metro Mirror

External StoragePROD HA

LUN level switching

External Storage

PROD HA/DR

Network

External Storage

Global Mirror

External Storage Replication

External Storage

PROD DR

Network

External Storage

LUN switch + Global Mirror

HA

PROD DR

Network

External Storage

Metro Global Mirror

HA

PROD HA

Network

Synchronous Geographic MirroringPROD HA/DR

Network

Asynchronous Geographic Mirroring

2 site Replication

2 site Replication

2 site HA + DR

IBM i replication

3 site Replication(DS8K only)

9

© Copyright IBM Corporation 2016.

• Additional benefits• Can switch to minimize OS upgrade outage• Design flexibility to minimize disruption from planned, unplanned, DR,

backups• Outage is a vary off and vary on of IASP (DB recovery steps of IPL)

Cluster Admin Domain

Sysbas Sysbas IASP

Sysbas

IASP

Lun level Switching

10

© Copyright IBM Corporation 2016. 10

Geographic Mirroring

• Synch• Any storage• Direct attached• SAN attached• VIOS or IBM i

Hosted storage

IBM i 7.1 / 7.2 PowerHA SystemMirror for i

MetroMirror

• Synch• DS8000• SVC• V7000• V3700• V5000• V840/v9000

FlashCopy

• DS8000• SVC• V7000• V5000• V3700• V840

SwitchedIASPs

• Internal or external storage

• IOA

End-to-EndSolution

NPIV and LUN Level Switching was added for SVC, Storwize and Native attach (No VIOS required in 1H 2013 – V840 and V9000 7.1 TR8 IBM i 7.2 TR2

GlobalMirror

• Asynch• DS8000• SVC• V7000• V5000• V3700• V840/V9000

Geographic Mirroring

• Asynch• Any storage• Direct, VIOS,

IBM i Hosted storage

LUN Level Switching

• DS8000• SVC• V7000• V5000• V3700• V840/V9000

Extended options for an IBM delivered, end-to-end solution for HA, DR and backups

IBM i Cluster Resource Services

DSCLI DS Command Line Interface

HA Switchable Resources - IBM i option 41 included

PowerHA SystemMirror for i(5770-HAS) – 7.1 /7.2

LabServices

ICSM

11

© Copyright IBM Corporation 2016.

IBM i Native Attach Storage and ResiliencyInternal SAS/SSD

(1) DS5000 DS8000 SVC V7000, V5000 V3700 V840 V9000

Non-IBM SAN (ie EMC) (2)

POWER/7/8 POWER/7/8 POWER7/8 POWER7 or 8 POWER7/8

PowerHA SystemMirror 7.1 or 7.2FlashCopy No No Yes Yes (3) No (Timefinder) (2)

Metro Mirror No No Yes Yes (3) No (SRDF) (2)

Global Mirror No No Yes Yes (3) No (SRDF) (2)

Switched IASP Yes Yes Yes Yes No

LUN Level Switching No No Yes (7.1) Yes No

Geographic Mirroring Yes Yes Yes Yes Yes

PowerHA SystemMirror 6.1 or 7.1 plus IASP Copy Services Manager (ICSM - formerly Advanced Copy Services (ACS))

FlashCopy No Yes Yes Yes No (Timefinder) (2)

Metro Mirror No Yes Yes Yes (3) No (SRDF) (2)

Global Mirror No Yes Yes Yes (3) No (SRDF) (2)

LUN Level Switching No No Yes (6.1) Yes (3) No

Metro/Global Mirror No No Yes No No

External Storage Full System Copy

FlashCopy No Yes Yes Yes (3) Yes (Timefinder)

Global Mirror No Yes Yes Yes (3) Yes (SRDF)

Metro Mirror No Yes Yes Yes (3) Yes (SRDF)

(1) SSD requires POWER6 or later. (2) EMC (DMX, VMAX) are not supported by PowerHA except with Geographic Mirror(3) Fabric Attach (SAN Switch) requires IBM i 7.1 TR6 + PTF’s, or Direct attach (no SAN switch – 4GB adapter only)V840 and V9000 (Flash) requires IBM i 7.1 TR8 or later (Note: 16GBs one both SVC and Storwize requires IBM i 7.1 TR10 or IBM i 7.2 TR2)Note. Native attach means the partition contains a SCSI, SAS or Fiber Channel card used to connect to the storage

For PowerHA -Storage is a key part of your HA/DR solution decision

12

© Copyright IBM Corporation 2016.

IBM i - VIOS Storage and Resiliency

PowerHA Supported

configuration options

DS5000 DS8000 XIV SVC / V7000 V5000 V3700 V840 V9000

POWER6/7/8

BladeCenter

POWER6/7/8

BladeCenter

POWER6/7/8

BladeCenter

POWER6/7/8

BladeCenter

PowerHA SystemMirror 7.1 or 7.2FlashCopy No Yes No Yes 2

Metro Mirror No Yes No Yes 4

Global Mirror No Yes No Yes 4

Switched IASP No No No No

LUN Level Switch No Yes No Yes 3

Geo’mirroring Yes Yes Yes Yes 2

PowerHA SystemMirror 6.1, 7.1 or 7.2 plus IASP Copy Services Manager (ICSM – formerly Advanced Copy Services (ACS))

FlashCopy Yes 1 Yes No Yes

Metro Mirror Yes 1 Yes No No

Global Mirror Yes 1 Yes No No

LUN Level Switch No Yes No No

External Storage Full System Copy

FlashCopy Yes Yes Yes Yes

Metro Mirror Yes Yes Yes Yes

Global Mirror Yes Yes Yes Yes

1 Requires NPIV capable fiber channel adapter / DS5000 NPIV support requires IBM i 7.1 TR22 SVC, V7000 requires IBM i 7.1 TR3 or later (vSCSI). 3 Requires NPIV capable fiber channel and IBM i 7.1 TR6 or later 4 V5000, V3700 support added with IBM i 7.1 TR6 and V840 / V9000 with TR8 or later

13

© Copyright IBM Corporation 2016.

A PowerHA Technology for Every Storage Type

InternalSAS/SSD DS8000

SVCV840,V9000

StorwizeXIV DS5000

Other Storage

Geographic Mirroring

Metro Mirror

Global Mirror

Metro Global Mirror

LUN switching

FlashCopy

HyperSwap ?

Recommend viewing IBM i POWER External Storage Support Matrix Summary – TechDoc - PRS4563 http://www-03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/PRS4563 andQuick reference - Storage options for PowerHA SystemMirror for IBM i – TechDoc - TD106243 http://www-304.ibm.com/jct03001c/support/techdocs/atsmastr.nsf/WebIndex/TD106243

14

© Copyright IBM Corporation 2016.

Technology for Every Outage Type

PowerHAunplanned switch

(<1 hr RTO)

PowerHAplanned switch (<30 min RTO)

LPM (0 RTO)

HyperSwap(0 RTO)

Planned OS outage (upgrade)

Unplanned OS outage (MSD)

Planned Server Outage (firmware upgrade, hardware upgrade)

Unplanned Server Outage

Planned Storage Outage

Unplanned Storage Outage

Site Outage - Disaster

15

© Copyright IBM Corporation 2016.

PowerHA and LPM Feature Comparison

PowerHA LPM

Live Partition Mobility between physical frames (IBM i 7.1 TR4) *

Server Workload Management**

Energy Management**

Hardware Maintenance

Software Maintenance

Automated failover upon System Failure (OS or HW)

Automated failover upon HW failure

Automated failover upon App failure

*~ 2 seconds of total interruption time** Require free system resources on target system

16

© Copyright IBM Corporation 2016.

PowerHA Price

PowerHA System Mirror for AIX and IBM i New price table effective March 1st 2016

TierEdition

small price/core medium price/core large price/core

standard - data center $2600 $3650 $4700

Enterprise – multi site $3400 $5250 $6800

PowerHA System Mirror for AIX and IBM i

TierEdition

small price/core medium price/core large price/core

standard - data center $2500 $3500 $4500

Enterprise – multi site $3250 $5000 $6500

License only the IBM i and PowerHA for cores needed on the target system, when using CBU. This could be as little as 1 core when using hardware based replication

17

© Copyright IBM Corporation 2016. 17

CBU for IBM i

Primary CBU

temporary entitlement transfer

• The CBU offering is used in high availability and disaster recovery deployments• Offering enables a customer to move workload between systems without fully redundant OS entitlements • Two year temporary keys eliminate redundancy for eligible LPPs• CBU designation available only upon purchase of a new box and must be registered to a qualified primary. • If a CBU is no longer affiliated with the original registering customer, it is not recognized as a CBU. • Registration process: client agrees to terms and conditions, CBU registration is validated, shipment is approved• CBU agreement requires that both the primary and CBU are owned by the same enterprise• With PowerHA you only need ONE core licensed on the CBU ! Generally not the case with logical replication

solutions. Logical replication solutions consume around 30% of CPWs ; all of those cores must be licensed with IBM i

18

© Copyright IBM Corporation 2016.

PowerHA Price Example…Economic Value - TCA

• PowerHA is priced per processor core used in the HA/DR cluster• Taking advantage of the CBU topology in the example topology:

• Assume S824 P8 Systems in the cluster:• 5 – IBM i OS and 5 PowerHA on production system• 1 – IBM i OS and 1 PowerHA on each (CBU A and CBU B) systems• IBM i …..savings: 7 licenses instead of 15 • PowerHA …savings: 7 licenses instead of 15• Temporary license keys for other IBM i LPPs on each of the CBUs

19

© Copyright IBM Corporation 2016.

19

Unix Linux Windows

Boot & OS

'C'

PGMs

'D''E'

'F'

DB

'G''H'

'I''J'

'K''L'

Object B Object C Object DObject A

Storage Management

IBM i

I/O

TIMI

Integrated or SAN

Storage Management Styles – IBM i Compared To...

System ASP

Page

Your IBM i solution is probably already delivering an enterprise class performance solution. Your future storage design should take this into account: for tier-1 storage, design for I/Os not capacity.

20

© Copyright IBM Corporation 2016.

PowerHA Basic Concepts

• PowerHA SystemMirror creates and manages a cluster topology• IASP volume group hosts the DB, IFS application data• Admin Domain manages the SYSBAS data

Admin domain

Application data

(IASP)

IASP -IndependentAuxiliary Storage Pool

21

© Copyright IBM Corporation 2016. 21

Independent Auxiliary Storage Pool

An IASP is: A set of disk units which contain a collection of user objects and the necessary system data

(e.g. storage management directories, object ownership, and authorization information) Independent of ASP1 (SYSBASE) – This enables the IASP to be taken offline or brought online

independent of system activity or other ASPs Provides an independent ‘Name Space’ for job and DB isolation

Can reside on internal or external disk – or a combination of both

Also known as – Independent Disk Pool

Availability of the IASP is controlled through varying on / off the associated device description and “attaching” jobs / threads to the IASP

Included in the base operating system

IASPs can contain: User defined file systems User libraries – some object types not supported

22

© Copyright IBM Corporation 2016.

Administration Domain (included in PowerHA)

– Changes on one node are propagated to others

• User profiles, System Values, Security, Key configuration data

PowerHA - Environment Resiliency Options

Synchronize non-IASP (SYSBAS) objects across systems in the cluster

Additional Monitored Resources (6.1)• Subsystem Descriptions (*SBSD)• Network Server Descriptions (*NWSD) of types

*WINDOWSNT, *IXSVR, and *ISCSI. • NWS Configurations (*NWSCFG)• NWSH Device Descriptions (*NWSHDEV)• NWS Storage Spaces (*NWSSTG)• Tape Device Descriptions (*TAPDEV)• Optical Device Descriptions (*OPTDEV)• Ethernet Line Descriptions (*ETHLIN)• Token-ring Line Descriptions (*TRNLIN)

Additional Monitored Resources (7.1)• Authorization lists (*AUTL)• Printer Device Descriptions (*PRTDEV)

Monitored Resources (5.4)• User profiles (*USRPRF)• Class (*CLS)• Job description (*JOBD)• ASP device description (*ASPDEV)• System values (*SYSVAL) • Network attributes (*NETA)• Environment variables (*ENVVAR)• TCP/IP Attributes (*TCPA)

LPAR-1 LPAR-1

IASP IASP

HA (target)PROD (source)

SYSBAS SYSBAS

23

© Copyright IBM Corporation 2016.

Admin Domain – Review, add or remove Monitored Remove

24

© Copyright IBM Corporation 2016.

What about Full System Replication

• Common Use: Need protection against (some) planned and unplanned outages for DR

• Less configuration (no IASPs) can be view as easier to implement• FlashCopy may be an option

Caveats• There is no active backup server – Target server is offline• Does not help with reducing software maintenance outage• Replicating OS and temporary storage (requires more bandwidth)• Failover is a full system IPL, not just data recovery steps (IASP vary

on)• Requires External storage

25

© Copyright IBM Corporation 2016.

iASP Benefits (vs full system replication)

Faster switching, no IPL No replicating OS, microcode, temp space Target system is online – just switch the data Better recovery – just data recovery steps Reduced bandwidth requirement Integrated with clustering BRMS networking integration Improved flexibility and masking planned outages Much simpler, automated switch process Consolidate workloads using separate iASPs Less impact for planned outages (PTF and OS upgrades)

26

© Copyright IBM Corporation 2016.

26

Components of a Resilient IBM i Environment

Tape Backup Tape Backup

1 1 1

23 5

4

1

7

7

75 5

66

1. PowerHA - Admin Domain - Plus Clustering Technology to facilitate and manage

2. Independent Auxiliary Storage Pools (IASP)

3. IASP Switching / LUN Level switching (requires a SAN)

4. Network capacity, performance and redundancy

5. Server or storage subsystem replication of IASP

6. Application availability

7. Fast or off-line backup of data

8. Proper planning, capacity, performance and scalability

6

8

7

Prod Local HA Remote HA or DR

27

© Copyright IBM Corporation 2016. 27

Clustering – at the heart of IBM i and PowerHA

Cluster: provides the communication infrastructure between systems and/or partitions Facilitates the execution of cluster events Simplified management, single point of control

Device Domain The device domain defines the cluster nodes which may share IASPs Each IASP is assigned an unique address space within the device domain Can’t allow two IASPs with the same address space to exist on the same IBM i node

OS and other system data

OS and other system data

clusterProduction

*SYSBAS

Device domain

Application data

IASP IASP

HA

*SYSBAS

28

© Copyright IBM Corporation 2016. 28

Cluster Resource Group

Cluster Resource Group (CRG) Defines which IBM i nodes are potential hosts for the IASP The recovery domain (list of nodes) is ordered (This determines replication direction) A switchable (takeover) IP address can also be defined and PowerHA will activate it on whichever node

is currently primary

Cluster resource group

Production

*SYSBAS

HA

*SYSBAS

Primary 1st Backup

IASP IASP

29

© Copyright IBM Corporation 2016. 29

Copy Description and ASP Session

Copy D The copy description describes one copy of the IASP Gives PowerHA all the information needed to access and control the IASP

Session: Describes the relationship between copy descriptions Will determine the type of replication from a system storage perspective PowerHA uses the session to control the replication

Copy description

HA

*SYSBAS

Copy description

sessionIASP IASP

Production

*SYSBAS

30

© Copyright IBM Corporation 2016. 30

Administrative Domain

Administrative Domain- List of cluster nodes to synchronize *SYSBAS objects- Independent of the data in the IASP

Monitored Resource Entries (MRE’s)- Objects in the Admin Domain being synchronized- Specific attributes of objects being synchronized

HA

*SYSBAS

IASP IASP

Production

*SYSBAS

cluster

Admin Domain

31

© Copyright IBM Corporation 2016. © Copyright IBM Corporation 2014

Geographic Mirroring – IBM

Synchronous or Asynchronous - long distances and DR• Two IBM i partitions (different “sites”)• One IASP, two copies (two sets of disks/LUNs)• Typically for internal storage but could have SAN at one side• Replication handled at IBM i SLIC storage management level

Internal Disks *

Internal Disks *GeoMirror

IBM iIBM i

PROD Target

Mirror Copy IASP

Production Copy IASP

High Availability (HA) and Disaster Recovery (DR) solutionReplication handled at SLIC Storage Management level, as opposed to O/S and remote journalingManaged via PowerHATrue asynchronous transmission option at IBM i 7.1Requires sufficient bandwidth between sites to maintain consistent copies and avoid auto-suspend issuesRecommended to have equal quantity, capacity and type of disks at both sites to maintain consistencyReplication ports use random ephemeral (1024+) TCP ports, not specific ports

32

© Copyright IBM Corporation 2016.

• Synchronous Geographic Mirroring – Local HAPROD (source) HA (target)

LPAR-1 LPAR-1

IASP IASP

SYSBAS SYSBAS

X X

= Main Storage

= Completed Write

• Synchronous write to target copy means the change must completed on target memory before it can complete on source

– Write at target can be synchronous or asynchronous (shown here as asynchronous)

• Very good RPO – changes are confirmed on target copy before source

• Requires local LAN like network performance: little latency, plenty of capacity

• Requires proper CPU, memory, disk planning for both source and targets to ensure performance expectations are met

X X

Your Network

IBM i Geographic Mirroring

IBM i 6.1 and up

I/O Adapter Cache I/O Adapter Cache

33

© Copyright IBM Corporation 2016.

Your Network

• Asynchronous Geographic Mirroring - long distances and DR

IBM i Geographic Mirroring IBM i 7.1 and up

PROD (source) HA (target)

LPAR-1 LPAR-1SYSBAS SYSBAS

X X

• Asynchronous write to target copy means network latency will not impact end user and job performance nearly as much as synchronous

– If the network delays writes to target copy, source storage pages wait at source

– Source memory pages will be paged out to disk, just as any other job in system

• Consumes additional source server CPU and memory compared to Synchronous Geographic Mirroring

• Available with PowerHA 7.1 and Enterprise Edition

MemoryPage Copy

= Main Storage

= Completed Write

IASP IASP

X XI/O Adapter Cache I/O Adapter Cache

34

© Copyright IBM Corporation 2016.

HA (target)

SYSBAS

PROD (source)

SYSBAS

IBM i Geographic Mirroring

LPAR-1 LPAR-1

IASP IASP

X XYour Network

• Limited use for on-line backups – Detach with Tracking

• Replication from source is suspended, changes are tracked

– Requires partial resynchronization once backups are completed

• No HA or DR failovers are possible until that re-sync has completed

– Will this meet your business requirements?

– By itself, can be a viable on-line backup solution, if full time HA/DR is not required. Otherwise, consider the latest version of Save-While-Active

• On-line backups at target site

No data replication during backupsPartial resynch

No HA or DR until resync completes

Detach with tracking

35

© Copyright IBM Corporation 2016. 35

LUN Level Switching – IBM External StorageExternal Storage using Switched IASPs technology for local HA

IASP

DS8000SVC/Storwize

PROD (source)

LPAR-1

HA (target)

LPAR-2

LUN Level Switching provides local High Availability

– IASP is switched via fiber connections between partitions

• Great for masking planned outages like release or application upgrades

– Or - IASP is switched via fiber connections between servers

• Great for masking planned and unplanned outages

– Local HA solution which can be used in conjunction with:

• Metro Mirror• Global Mirror• FlashCopy

DS8000 with IBM i 7.1 (or IBM i 6.1 with ACS supports LUN Level Switching) IBM i 7.1 TR6 adds NPIV support or direct attach for SVC, Storwize (V7000, V5000 & V3700)IBM i 7.1 TR8 or 7.2 TR2 adds supports for V840 and V9000 Flash

36

© Copyright IBM Corporation 2016.

LUN-Level Switching

• Local High Availability (HA) solution

• Automated failover through PowerHA, at v 7.1

• Managed via PowerHA (DS8K,SVC,Storwize)

• Limited amount of disk required

• Single copy of data

• Not a Disaster Recovery (DR) Solution

• Supported on DS8000, SVC, V7000, V5000, V3700 and V840/V9000

37

© Copyright IBM Corporation 2016.

PowerHA - Metro Mirror

PowerHA integrates storage-based replication with IBM i clustering technology–Synchronous replication ensures new production data is written to target first–Excellent RPO and very good RTO–An IASP based solution–DS8000, SVC/Storwize/V840 and V9000

• Direct attach, VIOS using VSCSI or NPIV–SVC, Storwize, V9000 and V840

• Requires 7.1 or later

ExternalStorage

HA

ExternalStorage

Local or campus Site

MetroMirror

Production

*SYSBASIASP

*SYSBAS

IASP

38

© Copyright IBM Corporation 2016. 38

Metro mirror

Synchronous mirroring: Metro Mirror receives a host update to the source volume, it completes the corresponding

update to the target volume. Guarantees data consistency by ensuring that a write operation that completes is received by

the host application after the update has been committed to the target storage unit and acknowledged by both the source and target storage units.

Metro Mirror supports a maximum distance of 300 km Delays in response times are proportional to the distance between the volumes. 100% of the

source data is available at the recovery when the copy operation ends.

Requires PowerHA SystemMirror Enterprise Edition (57xx-HAS Option 1) A Cluster with at least two nodes in a Device Domain A Device - Cluster Resource Group (CRG) with the two nodes in the Recovery Domain An IASP using SAN-based storage that are the preferred source volumes Identical number and size of SAN-based storage volumes that are the preferred target

(preferably in a different storage unit than the preferred source volumes) Copy Descriptions for the source and target volumes An ASP Session managing the metro mirror relationship

39

© Copyright IBM Corporation 2016.

Global Mirror Continuous Copy (SVC Storwize)

• Asynchronous copy; Writes sent sequentially in order• Requires sufficient bandwidth; Size for peak I/O workload• Better suited for fiber replication, not IP replication• Nearly unlimited distance; Maximum 80ms round-trip

PROD BCKP

Production Copy

Mirror Copy

Global Mirror

IBM i IBM i

SVC Storwize SVC Storwize

40

© Copyright IBM Corporation 2016.

Global Mirror + Change Volumes (SVC Storwize)

• Change volumes hold point-in-time copy that are changed during cycling mode – FlashCopy is performed

• Change volumes required at both source and target site (typically thin-provisioned)

• Requires 4 “sets” of disks/LUNs (two source, two target)

PROD BCKP

Production Copy

Mirror Copy

Global Mirror

IBM i IBM i

SVC Storwize SVC Storwize

Change Volume (FlashCopy) Change Volume

(FlashCopy)

41

© Copyright IBM Corporation 2016.

Global Mirror + Change Volumes (SVC Storwize)

• Benefits• High Availability (HA) and Disaster Recovery (DR) solution

• Replication handled at storage level, as opposed to IBM I

• Managed via PowerHA for SVC Storwize

• Does not require bandwidth to meet peaks

• Unlimited distance, up to 80ms round trip latency

• Recommended when performance is poor with metro mirror or global mirror continuous copy on SVC Storwize

• Additional disk capacity required for change volumes (auxiliary volumes)

42

© Copyright IBM Corporation 2016. 42

PowerHA DS8000 Global Mirror

Prod LPAR1

Local Site DS8000

IASP

DR Site

DR

DS8000

IASP StandardFlashcopy

ConsistencyGroup

Symmetrical – Enables reverse replication: required for role swap capability

StandardFlashcopy

ConsistencyGroup

Global MirrorGlobal Mirror

– Asynchronous replication allows for long distances– DS8000 (Direct attach or VIOS using NPIV)

Add copy descriptions for the volumes present in the global mirror environment Preferred source Preferred target Target consistency group flash volumes Source consistency group flash volumes for symmetric global mirror

– SVC and Storwize offers two async replications technologies (standard or Change Volume)

43

© Copyright IBM Corporation 2016.

Metro-Global Mirror

• Three IBM i partitions (three “sites”)• One IASP, three full copies, plus one additional set of disks/LUNs for

Consistency Group (CG) FlashCopy• Requires DS8000 external storage and TPC-R

IBM i IBM i IBM i

DS8800 DS8800 DS8800

MetroMirror Global Copy

PROD BCKPHA BCKPDR

Production Copy IASP

MMir Target GMir Source

PPRC Target Copy

CG Copy

44

© Copyright IBM Corporation 2016.

PowerHA Express Edition - Full System HyperSwap

First release (7.2) provides support for DS8000 HyperSwap in full system replication environments (SVC and Storwize with firmware 7.5 in testing)

• HyperSwap by itself is a hardware availability solution• ‘Zero’ downtime switch for storage planned and unplanned outages• Single partition solution, although can be combined LPM• Not a disaster recovery solution• No protection against software planned or unplanned outages

• Once configured, HyperSwap switch will occur automatically in the case of a DS8K failure, or can be triggered manually before a planned outage

Prod

MetroMirror

SYSBAS SYSBAS

Prod

MetroMirrorSYSBAS SYSBAS

LPM 2nd

server

45

© Copyright IBM Corporation 2016.

Next step… HyperSwap Stage 2 (7.2 future)

• IASP replication plus HyperSwap• HyperSwap for ‘zero’ downtime storage planned and unplanned outages• Live partition mobility for ‘zero’ downtime planned firmware, POWER outages• Challenge – two way replication (2 copies of Sysbase per PowerHA node) • PowerHA vary off/on of IASP for OS planned outage, disaster recovery

MetroMirror

Prod

*SYSBAS

IASP

DS8-A

IASP

DS8-B

*SYSBAS HA

*SYSBAS *SYSBAS

MetroMirror

MetroMirror

Current Production environment

Storage outage

46

© Copyright IBM Corporation 2016.

Server outage

• PowerHA Cluster detected failure – initiate Fail-Over to HA Target

• PowerHA vary off/on of IASP

• This will cover OS planned or unplanned outage, disaster recovery

MetroMirror

Prod

*SYSBAS

IASP

DS-A

IASP

DS-B

*SYSBASHA

*SYSBAS *SYSBASMetroMirror

MetroMirror

47

© Copyright IBM Corporation 2016. 47

Copy Services Copy Services is a real-time remote copy technique that mirrors a primary set of logical volumes (that are

being updated by applications) onto a secondary set of logical volumes. The secondary volumes can either be in the same or a different IBM System Storage Unit than the

primary set of volumes. Copy Services is a hardware solution, thus it is application independent. The copy function occurs at the

storage subsystem level, the application does not know of its existence.

Copy Services

System Storage

PowerHA

Integrates the storage system copy services functions with IBM i clustering technology to provide an integrated solution

– Copy Services technology does the replication of the IASP– Cluster technology monitors the health of the IBM i nodes as well as the health of the

copy services replication– PowerHA controls the direction of the replication within the storage server

48

© Copyright IBM Corporation 2016. 48

Flashcopy options

• IBM i can leverage FlashCopy to:

Create a copy of an IASP for backup

Create a full system copy for backup

• Recommendation is:

• Vary off the IASP or power down the system before taking the Flash

• Known as a “cold Flash” is the best way to guarantee complete data integrity

• IBM i 6.1 added support for the “Quiesce” of IASP

• Known as a “warm Flash”

• suspends transactions & operations to ensure that as much in-flight data as possible is written to disk

• Places transactions at database boundaries if possible Best when used with applications running commitment control

• Requires a ‘recovery vary-on’ of the IASP.

• 7.1 and later supports Quiesce with VIOS storage pools

49

© Copyright IBM Corporation 2016.

FlashCopy Multiple IASPs to Single Target

• Support added at PowerHA 7.2 for multiple source to 1 target

• Provides ability to create point-in-time “copies” of DIFFERENT production sets of data

• Eliminates scheduled outage time for production backups• Multiple backups can be staggered back-to-back

• Limited amount of disk required – saves on disk space, sharing same target LUNs

• Can be managed through PowerHA or ICSM

• Single target partition has only one “copy” at a time

• Multiple partitions involved

50

© Copyright IBM Corporation 2016.

IBM I 7.2 with PowerHA - IASP assignment

Enables use of one partition to save multiple production environmentsAllows attachment of an IASP to a partition not in the cluster device domainOnly one IASP can be attached to the partition at a timeEliminate dedicated flash partitions per clusterUniversal FlashCopy target node

Save LPAR

*SYSBAS

Tape Backup

FlashCopy targets

IASP Production cluster 1

IASP Production cluster 2

IASP Production cluster 3

51

© Copyright IBM Corporation 2016.

What is Active Partition Mobility?

• Active Partition Mobility is the actual movement of a running LPAR from one physical machine to another without disrupting* the operation of the OS and applications running in that LPAR.

• Applicability• Workload consolidation (e.g. many to one)• Workload balancing (e.g. move to larger system)• Workload mobility to newer systems.• Planned CEC outages for maintenance/upgrades• Impending CEC outages (e.g. hardware warning received)

* Applications may see a short (~2 second) blip.

52

© Copyright IBM Corporation 2016.

What is Inactive Partition Mobility?

• Inactive Partition Mobility transfers a partition that is logically ‘powered off’ (not running) from one system to another.

• Subject to fewer compatibility restrictions than active partition mobility because the OS goes through the boot process on the destination.

53

© Copyright IBM Corporation 2016.

What Active Partition Mobility Isn’t

• It is not a replacement for PowerHA.• It is not automatic.• LPARs cannot be migrated from failed CECs (Remote/Restart)• Failed OS’s cannot be dynamically migrated.

• It is not a Disaster Recovery Solution.• For the same reasons as above.• Mobility across long distances not supported.• Additional SAN and LAN considerations to treat.

• It is not an OS update solutions• LPM can help with server firmware updates by evacuating a server but

doesn’t help with OS upgrades

54

© Copyright IBM Corporation 2016.

LPM Requirements & Planning

• Source and destination must be mobility capable and compatible.

• Enhanced hardware virtualization capabilities (PowerVM Enterprise Edition)

• Identical or compatible processors (P7 and/or P8)• Compatible firmware levels (Power7 or Power8)

• Source and destination must be same subnet• Matching Processor Compatibility Mode

• All resources (CPU, Memory, IO adapters) must be virtualized prior to migration.

• (VIOS) Hypervisor will handle CPU and Memory automaticallyas required Virtual IO adapters are pre-configured, and SAN-attached disks accessed through Virtual IO Server IBM i unique, must set Restricted IO in properties

• Source and destination VIOS must have symmetrical access to the partition’s disks.

• e.g. no internal or VIOS LVM-based disks.

• OS is migration enabled/aware.• IBM i 7.1 TR4 PTF group –SF99707 level 4 or later initial

support• IBM i 7.1 TR8 or IBM i 7.2 Recommended • Certain tools/middleware can benefit from being migration aware

also.

SAN

LAN

BootPaging

Application Data

LPAR

HMC

55

© Copyright IBM Corporation 2016. 55

PowerHA SystemMirror – combining Technologies

Example: LUN Level Switching on DS8000 or SVC/Storwize, plus remote replication for a second copy of the data at DR location. LPM is possible with VIOS, can evacuate a node for server maintenance/hardware changes etc,

DR (target)

LPAR-4

DS8000SVC/Storwize

IASP

IASP

Metro or Global MirrorOr GeoMirror

Remote HA or DR – On-line backups

PROD (source)

IASP

DS8000SVC/Storwize

FlashCopy

BU - LPAR-2

On-line backups

PRODLPAR-1

VIOS LPAR1

LUN Level Switching – Local HA (for planned or unplanned) or LPM for work load balancing or planned server maintenance.

Local HALPAR-3 VIOS LPAR3

VIOS LPAR2

FlashCopy

BU - LPAR-5

Local HA &

Server A

Server B

Server CPRODLPAR-1

LPM LUN level Switching

56

© Copyright IBM Corporation 2016.

Summary

- Single-site solutions can provide some High Availability (HA) and may save $$, but cannot provide Disaster Recovery (DR)- Consider internal vs external storage (and what type of external storage), depending on which solution you choose- Consider connectivity options (Direct-attached, SAN-attached, VIOS, vSCSI, N-PIV) and number of host connections, when deciding which solutions can or cannot be implemented- Consider the amount of disk needed for multiple copies (Mirror Copy LUNs, FlashCopy LUNs, Consistency Group LUNs, Change Volume LUNs)- Consider BANDWIDTH requirements between sites to ensure efficient delivery of packets on mirroring solutions

57

© Copyright IBM Corporation 2016.

57

DEMO

58

© Copyright IBM Corporation 2016.

58

Demo configuration

i139

GeoMirror

8205-E6C (740)

SYSBASE

P11

P11 – Primary production partition running on 740

I139 – Geographic Mirror target running on Power 780

Application – IP address 9.5.101.112 - DNS name ‘GMTAKEOVER’

IASP

9179-MHDSYSBASE

IASP

59

© Copyright IBM Corporation 2016.

59

60

© Copyright IBM Corporation 2016.

60

DEMOV7000 Metro Mirror

61

© Copyright IBM Corporation 2016.

Demo configuration

PFHAbkup

V7000 - RATSPFV7

IASP

IASPMetro Mirror

8205-E6C (740)

IASP

FlashCopyi015raw

P14HAprdVIOS1-740

VIOS2-740

VIOS1

VIOS2

VIOS1-740

FlashCopy

P14HAprd – Primary production partition running on 740

i015HAraw – Flashcopy target partition for tape backup running on same 740 and same V7000

PFHAbkup – Metro Mirror target running on P8 S824

Application – IP address 9.5.101.158 - DNS name ‘MMTAKEOVER’

IASP

VIOS2-740

S824

V7000 - RATSV701

62

© Copyright IBM Corporation 2016.

62

Demo – LAB setup notesObject Name Name Name Name

Systems P14HAprd PFHAbkup I015raw (for flashcopy)

Unavailable at this time

Cluster name MMCLUSTER MMCLUSTER MMCLUSTER MMCLUSTERIASP – Metro Mirror

MMIASP MMIASP MMIASP MMIASP

CRG - Cluster Resource Group

SVC_MM_CRG SVC_MM_CRG SVC_MM_CRG SVC_MM_CRG

Device Domain MMCLUSTER MMCLUSTER MMCLUSTER MMCLUSTERRecovery Domain Site name

SITE1 SITE2

Admin Domain MMCAD MMCAD MMCAD MMCAD

IP addresses –Metro Mirror

9.5.101.108 9.5.101.156 9.5.101.177 9.5.101.181

Copy Descriptions SVC_MM_S SVC_MM_TASP Sessions MM_ASP_SSN MetroMirror MM_ASP_SSN MetroMirror

Userid web user MMPWRHA MMPWRHA MMPWRHA MMPWRHAUserid admin PHADEMO PHADEMO PHADEMO PHADEMO Password L0GIN4ME L0GIN4ME L0GIN4ME L0GIN4MEJOBD QGPL/MMPWRHA QGPL/MMPWRHA MMPWRHA MMPWRHA

URL TakeOver IP http//9.5.101.158:15000/ha DNSname - MMTAKEOVER

http//9.5.101.158:15000/ha DNSname - MMTAKEOVER

SVC (V7000) IP address

RATSV701 (9.5.101.120) RATSPFV7 (9.5.101.50) RATSV701 (9.5.101.120)

RATSPFV7 (9.5.101.50)

V7000 Vol IDs 205 – 210 29 – 34 229 – 234 41 - 46

Application port 15000 150000Application name hademo4 context root /ha hademo4 context root /haDB CONNECT name

MYCONN1 9.5.101.158 MYCONN1 9.5.101.158

Library in MMIASP phademo phademoIFS /home/demouser/hademo4.war /home/demouser/hademo4.war

DNS NAME = MMTAKEOVER

63

© Copyright IBM Corporation 2016.

Publications and Links (1 of 2)• PowerHA Wiki

• www.ibm.com/developerworks/ibmi/ha/• IBM PowerHA SystemMirror for i - Performance Information

• http://ibmurl.hursley.ibm.com/35XN• PowerHA solutions

• www.ibm.com/systems/power/software/availabilityRedbooks at www.redbooks.ibm.com

• PowerHA SystemMirror for IBM i Cookbook - SG24-7994-00 (Feb 2012)

• Implementing PowerHA for IBM i - SG24-7405-00 (Nov 2008)• IBM i 6.1 Independent ASPs: A Guide to Quick Implementation of Independent ASPs -

SG24-7811• Implementing SAP Applications on the IBM System i with IBM i5/OS - SG24-7166

Publications• Implementing high availability

• http://publib.boulder.ibm.com/infocenter/iseries/v7r1m0/topic/rzaig/rzaig.pdf• Resiliency on Power Systems

• www-03.ibm.com/systems/p/hardware/whitepapers/power6_availability.html• IBM i Virtualization and Open Storage Read-me First

• http://www-03.ibm.com/systems/resources/systems_i_Virtualization_Open_Storage.pdf

64

© Copyright IBM Corporation 2016.

Publications and Links (2 of 2)

PowerHA System Mirror for IBM i – Data Sheet• http://public.dhe.ibm.com/common/ssi/ecm/en/pod03024usen/POD03024USEN.PDF

Impact and Cost of Downtime Tools• IBM: http://www-935.ibm.com/services/us/bcrs/self-assessment/

• GIAC BIA overview: www.giac.org/resources/whitepaper/planning/122.php

Advanced Copy Services for PowerHA• http://www-03.ibm.com/systems/resources/systems_services_labservices_stgls_advcpysvc_powerha_10.pdf

• http://www.ibm.com/systems/services/labservices • Email: [email protected]

Business impact analysis (service)• www-935.ibm.com/services/us/index.wss/offering/bcrs/a1000260

IBM Risk Assessment Tool– www-935.ibm.com/services/us/gts/managing_risk/self_assessment/index.html