gse gdps virtual applianceconferences.gse.org.uk/attachments/presentations/aoqf7b_144650000… ·...
TRANSCRIPT
GDPS Virtual Appliance Sim Schindel
Executive IT specialist
GDPS Design & Development
IBM
1
Agenda
What is GDPS?
What is GDPS Multiplatform Resiliency for z Systems (xDR)?
What is the GDPS Virtual Appliance?
2
What is GDPS?What is GDPS?
3
Why GDPS?
How much interruption can your business tolerate?
Key questions you need to ask yourself:y q y y
Are you a global enterprise that operates across time zones and can no longer afford an “off-hours” window?
Does your business require continuous availability?Does your business require continuous availability? What is the cost of one hour of downtime during core business hours? Do you need an automated solution to meet your business continuity
requirements?
To determine your business continuity requirements:
Disaster recovery:Restore business after an unplanned compound failure
Standby(DR)
– Restore business after an unplanned compound failure High availability:
– Meet service availability objectives, for example, 99.9 percent availability or 8.8 hours of downtime a year
Continuous availability:
4
y– No downtime (planned or unplanned)
Active/Active(CA)
What is GDPS?
Manages application and data availability in and across sites – Monitors systems and disk subsystems– Manages planned and unplanned activities
System/disk maintenance / failure Site maintenance / failure
– Manages planned and unplanned activities across platforms– Manages planned and unplanned activities across platforms Multiplatform Resiliency for z Systems (xDR for Linux on z Systems) Distributed Cluster Management (DCM)
Builds on proven high availability technologies– Clustering– Remote copy – Automation
Integrated / Automated solution Easy to use interface
– Panel/GUI interface
5
– Policy based commands
GDPS-managed disk-based sync / async copy technologies
Metro Mirror (PPRC) Synchronous remote data mirroring Metropolitan Distance ( C)1 4
Sys zSys zz/OSz/OS
UNIXUNIXWinWin
Metropolitan Distance IBM z Systems and distributed data Multi-target PPRC support GDPS provides data consistency
Metro Mirror (PPRC)1
2
3
z/OS Global Mirror (XRC) Asynchronous remote data mirroring Unlimited distance support
IBM S t d t
z/OS Global Mirror(XRC)
1 2
SDM
3 4 IBM z Systems data System Data Mover (SDM) provides data
consistency
Gl b l MiGlobal Mirror Asynchronous remote data mirroring Unlimited distance support IBM z Systems and distributed data
UNIXUNIXWinWin
Global Mirror1 2
63
Sys zSys zz/OSz/OS
6
Global Mirror provides data consistency
4
FCA
B
C5
Multiple GDPS products meeting various requirements for Local/Metro/Remote Continuous Availability and/or out-of-Local/Metro/Remote Continuous Availability and/or out ofregion Disaster Recovery
Continuous Disaster Recovery CA Regionally and CA, DR, & Cross-ContinuousContinuous Availability and
entry-level Disaster Recovery
GDPS/PPRC HM
Disaster Recovery Extended Distance
GDPS/GM & GDPS/XRC
CA Regionally and Disaster Recovery Extended Distance
GDPS/MGM & GDPS/MzGM
CA, DR, & Crosssite Workload
Balancing Extended Distance
GDPS/Active-Active
Continuous Availability with
DR within Metropolitan Region
GDPS/PPRC & GDPS/MTMM
Si l D t C t
RPO=0 [RTO secs]for disk only
T D t C t
RPO secs, RTO<1h
T /Th /F
RPO=0,RTO mins/<1h& RPO secs, RTO<1h
T A ti
RPO secs, RTO secsGDPS Virtual ApplianceRPO=0
RTO mins / RTO<1h
T D t C tSingle Data Center(one or two sites)
Applications remain active
Two Data Centers
Rapid Systems DR with “seconds” of
data loss
Two/Three/Four Data Centers
High availability for site disasters
Two or more Active Data Centers
Automatic workload switch in seconds;
seconds of data
Two Data Centers(2 server sites, 2 or 3 disk
locations)
Systems remain active
Continuous access to data in
the event of a storage outage
Disaster Recoveryfor out of region
interruptions
Disaster recovery for regional disasters
seconds of data lossMulti-site workloads
can withstand site and/or storage
failures
RPO – recovery point objective RTO – recovery time objective
GDPS/PPRC for two sites – Metropolitan distance continuous (C ) ( )availability (CA) and disaster recovery (DR) solution
Continuous Availability / Disaster Recovery within Provides sysplex, system and server managementDisaster Recovery within
a Metropolitan Region
Two Data Centers
Systems remain active
y p y g Simplifies and streamlines data replication management Manages remote copy environment using HyperSwap function
and keeps data available for operating systems and applications (extends Parallel Sysplex CA function to disk data)Systems remain active
Multi-site workloads can withstand site and/or
storage failures
(extends Parallel Sysplex CA function to disk data) Facilitates faster recovery time for planned and unplanned
outages Ensures successful recovery via automated processes
zOSSysplex
Linux /zVM
y p Enhances data consistency across all secondary volumes for
both z Systems and distributed systems Enables high availability and DR for Linux on z Systems
environments
GDPS/PPRC active/active,
active/standby configs
environments Leverages Distributed Cluster Management (DCM) to interface
with distributed environments to provide an enterprise-level disaster recovery solution
8
active/standby configs
Mirroring Failure vs. Primary I/O Failure
Primary Failure (HW):
Can not write to primary disk
PPRC Mirroring Failure: Can write to primary disk
C t i t d Captured by z/OS on 'HyperSwap-
ready' systems
z/OS notifies GDPS via ENF
Can not mirror to secondary Can be for different reasons
– Link interfaces on primary / secondary disk subsystem broken z/OS notifies GDPS via ENF
If PRIMARYFAILURE=SWAP,xxx GDPS performs Swap else GDPS performs Freeze
– DWDM broken– Fiber broken– Secondary disk broken
GDPS will perform Freeze and follow on action based on PPRCFAILURE policy performs Freeze
The solution for primary failures is HyperSwap
action based on PPRCFAILURE policy option The GDPS solution for mirroring
problems is Freeze automation
PPRCPPRC
9
SP
Primary Disk Secondary Disk
P S
Primary Disk Secondary Disk
What is GDPS MultiplatformWhat is GDPS Multiplatform Resiliency for z Systems (xDR)?
10
GDPS xDR: Linux Guests and Native Linux on IBM z Systems – Continuous Data Availability
site 1 site 2
SA MPProxy
SA MPMaster
SA MP SA MP
Multiple Clusters
z/OS SysplexGDPS
z/VMBackupLPAR
Linux BackupLPAR
Linux[SUSE]
z/VMLinux Guest[RHEL, SUSE]
z/OS SysplexGDPS
K1 K2
site-1 site-2z/OS spanned Sysplex
[active/active or active/standby]
yNode Node Node Node
Multiple Linux clusters containing multiple nodes butno proxy partition
K1 K2
PPP
ECKD / FBA ECKD ECKD
S S S
Coordinated HyperSwap – z/OS, z/VM with its guests, and native Linux Graceful shutdown and startup (re-IPL in place) of Linux on z Systems clusters or nodes
GDPS-managed PPRC
Coordinated takeover – recovery from a Linux node or cluster failure Multiple Linux clusters are supported, as are multiple z/VM systems & Linux LPARs All members of a cluster must run under same z/VM system or SSI or in same Linux
LPAR
11
Coordinated recovery for planned and unplanned events
GDPS/PPRC Multiplatform Resiliency for z Systems (xDR) A hit t l B ildi Bl k– Architectural Building Blocks
Business Continuity solution for z/OS and Linux applications on z Systems Leverage existing, proven solutions
GDPS
SA MP SA zOS– GDPS
– IBM Tivoli System Automation for z/OS
– IBM Tivoli System Automation for NetView
Linux on z
Linux on zGuests
Other Multiplatform (SA MP)
Coordinated cross platform business resiliency for operating systems (OS) running on System z hardware
z/OS
Linux on z Native
z/VM
OtherGuests
running on System z hardware
Integration point of z/OS, z/VM and Linux on z Systems
System z Hardware
12
System and data management across z/OS & Linux on z Systems
GDPS xDR Support
xDR extends GDPS to support not only z/OS but also Linux on z Systems
Di k d iGDPSK s s– Disk error detection
– Heartbeat for sanity checks– Re-IPL in place x
K-sys
– Coordinated Site Takeover– Coordinated Freeze&HyperSwap
Linux running as guest on z/VM (xDR on z/VM)
LinuxSA MP
LinuxSA MP
LinuxS
A MP
xDR
Proxy
SA M
P
Linux running as guest on z/VM (xDR on z/VM)– Linux Distributions - RHEL, SUSE– Disk Types – ECKD (FB not supported) z/VM
System z
CP
Linux running native in LPAR (xDR native)– Linux Distributions SUSE SLES11– Disk Types – ECKD
P S
y
13
P S
Note: xDR is not a product – to implement xDR the following products are required: GDPS, SA MP, Linux, VM if Linux is running on z/VM
xDR Functional Scope
GDPSK-sys
Logical layer for disks Disk error detection HyperSwap
LinSA
ProSA
HyperSwap Re-IPL in place Integration with SA MP
Shutdown & startup of systems and clusters
STOPSTART
STOP
STARTSTOPSTARTLinux
SA MPLinux
SA MP
nuxM
P
oxyM
P – Shutdown & startup of systems and clusters– Start of resources/applications after IPL– Stop of resources/applications before
shutdownCP
START
z/VM
System z Integration with z/VM
– Coordinated IPL/shutdown of z/VM1. Stop SA MP Clusters
STOPSTART
PS
2. Stop z/VM guests3. Stop z/VM
14
Coordinated recovery for planned and unplanned events
xDR Functional Scope (cont)
Coordinated recovery from a Linux on z Systems node or cluster failureGDPS awareness of node/cluster failure through heartbeat monitoring with managed– GDPS awareness of node/cluster failure through heartbeat monitoring with managed nodes
– GDPS can automatically re-IPL a failing node or all nodes in the failing cluster – Data consistency across Linux on z Systems and/or z/VM
Disk Subsystem maintenance (planned actions)– Swap z/VM and its guests from primary to secondary PPRC devices transparently
Disk Subsystem failure (unplanned actions)S di ti l /VM d it t f i t d PPRC d i– Swap non-disruptively z/VM and its guests from primary to secondary PPRC devices following a failure that GDPS recognizes as a HyperSwap trigger
Single point of control to manage disk mirroring configuration Coordinated recovery for other guests besides Linux on z SystemsCoordinated recovery for other guests besides Linux on z Systems
– TPF, VSE
15
Providing CA for z/OS and Linux on z Systems z/VM Guests- Disk and site failure scenarios
GDPS Site Takeover
CBU
SAP SAP PrxySAP SAP
SAMP
SAMP
SAMP
SAMP
SAMP
NetViewSA zOS
SAP
NetViewSA zOS
SAP
NetViewSA zOS
GDPS
OOCoD
d bl d bl
LPAR1z/VM
SAP
L
SAP
U
Prxy
X
SAP
I
SAP
N
LPAR2
DBserver
LPAR3
DBserver
LPAR4
GDPSK-sys
LPAR5 LPAR6
expandableworkload
expandableworkload
z/OS Sysplex
CKDCKD
HyperSwap
/OS Sysp e
site 1 site 2
CKDCKDMetro Mirror
Multiple Linux on z Systems clusters are supported, as are multiple z/VM systems All the members of a managed cluster must run under the same z/VM system
16
Coordinated recovery across z/OS, z/VM and Linux Guests
xDR High Availability dual Proxy configurationGDPS/PPRC xDR guest
Dual proxy cluster K-sysalternate
K-sysmaster
– One proxy on Site1 disk
– One proxy on Site2 disk
Master Proxy on disk alongside SA MPClustery g
PPRC secondary disk– Automatic proxy switch
High Availability configuration
LinuxSA MP
ProdSys
ProdSys Proxy Proxy
site 2/VM
master
g y g– Proxy & Proxy disk no longer a
single point of failure Requires SA MP 3.2.2.2 proxy proxy
site 2
site 1
z/VM
P SPPRC
17
Proxy Master role is switched automatically when PPRC diskis switched (or recovered) or when the Master proxy fails
xDR Configuration Verification Programs (CVP)
xDR has many components– Each component requires specific setup
GDPS impacted if setup incorrect
CVP validates setup– Reports what is setup incorrectly
GDPS
SA MP SA zOS– Helps ensure installation and
customization tasks performed correctly
Manually initiated
NetView
Linux on z N i
Linux on zGuests
OtherG
y– Separate programs for z/OS and Linux
Can be rerun when environment changes– Ensure setup is not broken by changes System z Hardware
z/OSNative
z/VM
Guests
Ensure setup is not broken by changes
Provided for both xDR Native & Guest environments
Note: xDR CVP requires SA MP 3.2.1.2
System z Hardware
18
Improved availability for xDR
xDR Monitor
Extensive checking of a large number of conditions that would prevent successful HyperSwap or impact GDPS operations for exampleHyperSwap or impact GDPS operations, for example,
– Channel paths to secondary disks online (Native & Guest)– Proxy system disk not PPRCed (Guest)– Minidisk caching is not enabled (Guest)– Minidisk caching is not enabled (Guest)
Alerts raised or certain ‘broken’ conditions autonomically fixed Identifies and flags xDR systems that are not HyperSwap-ready
Prevents planned HyperSwap which would cause not ready systems from getting reset– Prevents planned HyperSwap which would cause not-ready systems from getting reset– Allows client to fix identified problems for fully successful HyperSwap
Provided for both xDR Native and Guest environments
19
Improved availability for xDR
What is the GDPS Virtual Appliance?What is the GDPS Virtual Appliance?
20
GDPS Virtual Appliance extends GDPS capabilities into z/VM and S /OSLinux on z Systems environments that do not have z/OS.
GDPS Virtual Appliance features:
Single point of control and automation
– Reduced need for highly specialized skills to handle recovery and planned site switches Li
nux
Linu
x
xDR
Pro
xy x
2
GDPSAppliance
Linu
x
Linu
x
xDR
Pro
xy x
2
Manages remote copy environment and keeps data available and consistent
Transparent disk switch viaSystem z
LPARLPAR
System z
LPAR
z/VM z/VM
Transparent disk switch via HyperSwap®
– planned/unplanned disk/site outages
Monitoring and automation z/VM + Linux
disks(primary)
PPRCz/VM + Linux
disks(secondary)– reliable and rapid recovery via
automated repeatable processes
Virtual Appliance requires:– General purpose engine (CP)
(primary)
Site 1 Site 2
(secondary)
21
General purpose engine (CP)– z/VM® and Linux on System z guests– SA MP– ECKD Disk
Takeover
GDPS Virtual Appliance LPAR
Runs the GDPS control code to manage the environment and coordinate HyperSwap
R i Requires:– 1 dedicated General Purpose engine: CP (can not run on IFL)– 2 GB Central Storage
4 3390 volumes4 3390 volumes– 1 mod 9– 3 mod 54– Fixed addresses 2030-2033
A t ll PPRC l ( i d d ) Access to all PPRC volumes (primary and secondary)
22
Deployment Manager system
Used to customize and deploy GDPS Virtual Appliance images
R Li t f th d t Runs as a Linux guest on one of the managed systems
Requires:– 1 GB Storage– 80 GB minimum size for root filesystem80 GB minimum size for root filesystem– Any supported Linux distribution– Additional rpm (gdps-appl-deployment-manager-1-0.s390x.rpm) available on GDPS
secure FTP site with images– Access to GDPS Appliance LPAR 2030 and 2033 devicesAccess to GDPS Appliance LPAR 2030 and 2033 devices
23
z/VM Managed Systems• Multiple z/VM systems supported• Each z/VM system must contain at
least two Linux guests running SA MP.• This is the GDPS “proxy” cluster
Each z/VM may also contain:• Each z/VM may also contain:• One or more additional Linux guests
running SA MP and some application workload.
• One or more additional Linux guests without SA MP and some application
Linu
x
Linu
x
xDR
Pro
xy x
2
GDPSAppliance
Linu
x
Linu
x
xDR
Pro
xy x
2
without SA MP and some application workload.
• One or more non-Linux guests (CMS, etc. except z/OS)
• TCP/IP Connections from GDPS Vi t l A li t
System z
LPARLPAR
System z
LPAR
z/VM z/VM
Virtual Appliance to:• Proxy guests (required)• Other Linux guests with SA MP
(optional)• SA MP Clusters:
z/VM + Linux disks
(primary)PPRC
z/VM + Linux disks
(secondary) • Proxy guests must be separate cluster• Group other Linux nodes into clusters,
as appropriate for efficient SA MP design.
(primary)
Site 1 Site 2
(secondary)
24
xDR for Linux on z/VM Software Requirements
Supported Linux Distributions
Supported Linux Distributions Version of System Automation
SUSE SLES 11 SA MP 4.1.0.1 (plus APAR) or higher
Red Hat 6 SA MP 4.1.0.1 (plus APAR) or higher
Separately priced SA MP feature xDR for Linux is required Separately priced SA MP feature xDR for Linux is required Latest fixpacks for Linux and SA MP Required VM Version
V5 4 or V6 2 or higher– V5.4, or V6.2 or higher– Check GDPS PSP bucklets for recommended maintenance
Supported Disk Types– DASD ECKD
25
DASD ECKD
Restrictions (compared to full function GDPS/PPRC xDR)
No support for disks defined in alternate subchannel set (MSS1)
N t f SSI d Li G t R l ti No support for SSI and Live Guest Relocation
No support for GDPS managed FlashCopy
26
Implementation Summary
x2
Appliance Deployment
Manager
GDPS Appliance
Image 1. Project kick off and technical training2. Setup GDPS Appliance LPAR meeting
Linu
x
Linu
x
xDR
Pro
xy x
GDPSAppliance
z/VM
p pp gappliance requirements
3. Define/Install Deployment Manager on Linux on z as a guest or native in LPAR meeting requirements
3 C li i t di k ibl b
System z
LPARLPAR
3. Copy appliance image to disk accessible by Deployment Manager
4. Implement and customize xDR on Managed z/VM systems
5 Run Deployment Manager to copy
Appliance
Disk
5. Run Deployment Manager to copy appliance image to one of its RES volumes
6. IPL LPAR from Appliance Disk7. Complete customization of GDPS Appliance
environment and policy including system,
Site 2environment and policy including system, processor, mirroring configuration, and script definitions
*** Red text denotes tasks performed primarily by the customer
Time to deploy: minutes vs days
customer
GDPS Virtual ApplianceTechnical Overview Session Operation
28
User Interface: CLI
Runs under USS on the Appliance
A d i P TTY t l t i W bUI P l Accessed via PuTTY telnet session or WebUI Panel
Used to perform administrative functions– Password changes– Manage GDPS Services (WebUI, CLI, NetView)Manage GDPS Services (WebUI, CLI, NetView)– Shut down the Appliance system (must IPL from HMC)– Collect logs and/or dumps for problem determination– Work with user files stored on the Appliance– Apply hot fix maintenanceApply hot fix maintenance
29
CLI commands gdps status|version
gdps passwd [<user>]
d h gdps show geoparm
gdps scripts (list|del <script name>|show <script name>|add <filename>|update <filename>)
gdps restart [webui|cli]
gdps shutdown
gdps update io config <filename> gdps update-io-config <filename>
gdps prepare-upgrade
gdps collect-logs [<charset>] [ftp PMR-number]
gdps collect-dumps <dumpname> [ftp] PMR-number
gdps hotfix <hotfix-package>
gdps rescue <rescue-package> <output-file> [ ftp pmr-number]
30
User Interface: WebUI
New browser based GUI
R W bS h i t th t i t t ith th GDPS N tVi Runs as a WebSphere instance that interacts with the GDPS NetView
Used to perform most day to day management functions– Planned Actions– Standard ActionsStandard Actions– PPRC Management– Script invocation– Status monitoring
31
WebUI Dashboard
32
WebUI SDF Panel
33
WebUI Manage Geoparm Panel
34
WebUI Manage Add SSID Wizard
35
WebUI LSS Pairs Panel
36
WebUI Pairs Panel
37
WebUI Standard Actions Panel
38
WebUI Standard Actions Panel – guest nodes
39
WebUI Planned Actions Panel
40
WebUI Administrative console Panel
41
Support Model
Support handled through normal Support Center PMR process.
d ll t l d d ll t d CLI d i t ith di ti d t gdps collect-logs and gdps collect-dump CLI commands assist with diagnostic data collection
– FTP parameter to optionally transmit the data driectly to ECUREP.
Rescue packages for special needs.Rescue packages for special needs.– Special automated jobs to perform whatever is required.– Built by Support/Development and sent via ECUREP.
Avoid engaging IBM Service to diagnose and resolve problems.
42
Service/Maintenance Model
Two GDPS Virtual Appliance images will be released per year.A il f h thi f h ill i k th N 1 GDPS/PPRC i / l– April refresh - this refresh will pick up the N-1 GDPS/PPRC version/release
along with maintenance.– October refresh - this refresh will pick up the N GDPS/PPRC version/release
along with new function and maintenance.g– Customers are advised to run on the N-1 release.
Hotfix packages for fixes that can not wait for the next release– If required, a cumulative hotfix package will be maintained for each image equ ed, a cu u at e ot pac age be a ta ed o eac age
release.– Availability will be announced via GDPS Newsletter.– Installed using CLI command.
Activated dynamically if possible– Activated dynamically if possible.
Maintenance refresh images– `Released if installation of maintenance via hotfix becomes more disruptive.
43
Summary
GDPS Virtual Appliance provides:– GDPS capabilities to zVM/Linux clients with no z/OS systems or skillsGDPS capabilities to zVM/Linux clients with no z/OS systems or skills– Single point of control and built-in automation reducing the need for highly
specialized skills to handle recovery and planned site switches• Intuitive GDPS WebUI
R t t d d t i t– Remote copy management and data consistency– HyperSwap® for transparent disk swap for planned/unplanned disk outages– Monitoring and automation to ensure reliable, repeatable and rapid recovery
via automated processesp
Based on GDPS/PPRC:• 17 years in the market with continually evolving functions/capabilities• Industry leading mainframe Continuous Availability & Disaster Recovery solution
44
Questions?
45
Additional Information Web sites:
GDPS www.ibm.com/systems/z/gdpsParallel Sysplex www.ibm.com/systems/z/psoBus Resiliency z www ibm com/systems/z/resiliencyBus Resiliency z www.ibm.com/systems/z/resiliencyBus Resiliency www.ibm.com/systems/business_resiliency System z www.ibm.com/systems/z/hardware
Storage www.ibm.com/systems/storageRedbooks®GDPS Family: An Introduction to Concepts and Capabilities
www.redbooks.ibm.com/abstracts/sg246374.html?Open GDPS Web Site White Papers and Presentations
GDPS: The Ultimate e business Availability Solution– GDPS: The Ultimate e-business Availability Solution – IBM Implementation Services for GDPS/Global Mirror – GDPS Business Continuity Solutions – Consistency Groups in a Nutshell
DS8000™ Data Replication– DS8000™ Data Replication – GDPS Solutions
e-mail: [email protected]
Dank uDutch Merci
FrenchGracias
Spanish
СпаcибоRussian
شكراً 감사합니다 děkuji
ध यवाद Hindi
شكرا Arabic
감사합니다 Korean
תודה רבה Hebrew
děkujiCzech
Teşekkür ederim
Thank YouObrigado
BrazilianPortuguese
DankonEsperanto 谢谢
Teşekkür ederimTurkish
TakDankeTrugarez
Breton
ありがとうございますJapanese
谢谢Chinese
DanishGerman
ขอบคุณThai
GrazieItalian நன்றி
Tamil
Tack så mycket Thaigo raibh maith agat
Gaelic
ac så yc eSwedish