PowerVM: HADR Solutions
VM Recovery Manager HA DR
Jes Kiran
Architect, VM Recovery Manager HA & DR
Agenda
• HADR solutions
• VMRM 1.5 features
• SAP HANA Deployment models
• Demo
© 2020 IBM Corporation 2
VM Recovery Manager HA & DR: Deployment Modes
ü Uniform interface to manage HA & DR
ü Consistent setup, configuration,
policy management for both HA & DR
Fig 1: VM Recovery Manager HA mode Fig 2: VM Recovery Manager DR mode
(Shared storage, Auto restart) (Replicated storage, Manual restart)
• HADR mode: Medium HADR protection
Site 2
Site 1 VM Restart – HA is done only on home site
– DR failover to backup site allows Business continuity w/o
Control System K-Sys HA
System 1
• HADRHA mode: Full HADR protection (2020)
System 2 System 3 Restart System 4
Restart (HA) Restart (DR) – HA is done both sites
(HA)
VM 1 Restarted – After DR failover to backup site HA resumes
VM 1 Restarted
VM 1 – Requires full capacity hardware on backup site
VM 1
Replication
Fig 3: VM Recovery Manager HADR mode or HADRHA mode
© 2020 IBM Corporation 3
VM Recovery Manager HA Capabilities
Error handling AIX Linux IBM i
• End to End HA management: • Application HA management: VM Agent Available Available Not available
– Protect against Host, VM, & Application failures – Monitor any application inside VM
Host Failure ✓ ✓ ✓
– Deep monitoring & Automated failover – Application start sequencing
VM Failure ✓ ✓
• VM Relationships: – Tiered recovery
Application Failure ✓ ✓
– Collocation & Anti-collocation, Host black list – HA agents for SAP, Oracle, & DB2
– Workgroup (VM Group) • Best Fit placement:
• Flex Capacity: – Based on spare capacity, policies
– Adjust capacity during failover to non-Home Host • Graphical management: § Deep integration with PowerVM
• Planned HA through LPM: – Deployment, health monitoring, § Power 7+, 8, 9 support
administration experiences § AIX, IBM i, Linux VMs
– Migrate to 1 to many VMs
– One click Host vacate or restore § Auto discovery of changes
§ Daily validation
© 2020 IBM Corporation 4
VM Recovery Manager HA: Application HA management
AppMon: Lightweight application monitoring framework
KSYS
Application HA administration from inside VM
HMC
ksysvmmgr [command] [Resource Type] [ Resource Name] [flags]
ü Register/add application ü Controls to customize health VIOS
ü Unregister app monitoring. # of restarts, scripts etc
ü Modify application config
ü Suspend/Resume HA mgmt. APP
ü Start/stop application Mon
ü Application health status
VM
Fig 2: App failure handling
Fig 1: KSYS to VM com
App Auto start Application relationships
ü No Auto start ü Parent-Child
ü Auto start after VM boot ü Start after HA Agents
ü Controlled by KSYS ü Stop after • SAP HANA (HSR) HA
• Oracle
• DB2
• Postgres
© 2020 IBM Corporation 5
Application HA management: Application start/stop orchestration
1. Sequence applications across VMs (Fig 1)
• Start/stop sequencing applications in the environment
K-SYS
2. Start or stop all applications in a VM or a Host AIX LPAR
• Easy to use interface to control applications in the environment
VM 1 VM 2 VM 3
APP 3
APP 1 APP 2 APP 4
Fig 1: Application orchestration
© 2020 IBM Corporation 6
What is VM Recovery Manager DR (GDR) for Power
VM Restart based DR: Simplified Disaster Recovery Solution for Power
A simplified way to manage DR
§ Automated Disaster Recovery management
§ Economics of eliminating hardware and software
resources on backup site
– Enterprise Pool support (optional)
§ Easier deployment for DR: unlike clustering or
middleware replication technologies
§ VM restart technology has no OS or middleware
dependencies
ü Support for IBM POWER7® , POWER8®, & POWER9® Systems ü Enterprise Pool support: DR site for less
ü Support for heterogeneous guest OSs ü Storage replication mgmt: EMC, SVC/Storwize. DS8K, Hitachi, XIV/A9000
– AIX ü Extensive validations
– Red Hat ü Pluggable framework for customization
– SUSE ü Easy to deploy: less than 10 steps to deploy
– Ubuntu
– IBM i
© 2020 IBM Corporation 7
VM Recovery Manager DR (GDR) Capabilities
• Easy to deploy & manage: 1 to N VMs • Priority Restarts: Priority based DR restarts
– Admin initiated automated DR (Planned/unplanned). – Assign High, Medium, Low priorities & DR restart
– Advanced validations. Site/Host/VM group failover
§ Deep integration with
PowerVM
§ Power 7, 8, 9 support
§ AIX, IBM i, Linux VMs
§ Auto discovery of
changes
• Failover Rehearsal: Deploy, Login, Test in Test VMs on § Daily validation
backup systems • Flex Capacity: Capacity adjusted DR restarts
– Bring up middleware. Perform read/write tests – Assign CPU and memory percentages for backup site
§ Resync with ease
– Test frequently to meet compliance
§ Replication management
for many storages
§ Graphical management
© 2020 IBM Corporation 8
Flexible DR management: M to N Server pairing support
Site 1 Site 2
(home,active) (backup) Site 1 Site 2
(home,active) (backup)
Fig 1: 1 to 1 Host Pairing model Fig 1: M to N Host Pairing model (Here 4 to 2) (2019)
• Hosts are paired one to one • Hosts paired many to many
– VM Failover is within the pair across sites – Optimized resource deployments
– Minimum granularity for DR is 1 host – Flexible DR model
– Con: Full capacity requirement – Ideal for cloud deployments
– Con: VM’s can not be optimally placed
© 2020 IBM Corporation 9
Flexible DR management: VM group based DR management
Site 1 Site 2
(home,active) (backup)
Server 1 Server 2 Server 3 Server 4 Server 5
Power
KSYS
VM
group 1
VM11 VM12 … VM1m VM21 VM22 … VM2n VM31 VM32 … VM3p
VM
group 2
VM25 … VM2o VM35 … VM3q
VM
Group 3
VM36 … VM3r
Benefits
• DR management at Work group/VM group granularity
– Share servers and manage multiple clients easily
– All the DR management operations at VM group level
ü Failover rehearsal of a VM group/Host group
ü Discovery/validation of a VM group
ü Planned or unplanned move at VM group level
© 2020 IBM Corporation 10
VM Recovery Manager HA & DR: 2020
VM Recovery Manager HA & DR
1.4 SP1 1.4 SP2 1.5
Mar 20 Jun 20 Nov 2020
• Usability fixes and • Selective Disk disable • Role based management
improvements (unmanage disk) • DR site HA management (HADRHA
• Graphical management deployment)
improvements • Scalability improvements
• DR activity reports • Improved events and notifications
• Events reset support • SR-IOV override support
• Migration improvements • Health monitoring improvements (local mode)
© 2020 IBM Corporation 11
SAP HADR with VMRM
© 2020 IBM Corporation 12
HA & DR Approaches : SAP HANA High Availability
Cold Restart HA solutions
KSYS
Cold
Restart
• Slower Recovery
Node 1 Node 1
Active Restarted • Cost Savings:
– One copy of storage
– Standby could have fewer CPU and memory
during normal times
Fig1: VM Restart HA
Warm Restart HA solutions
HSR
Node 1 Node 2 • Faster Recovery
Primary VM Restart Secondary • Expensive:
– Twice the storage, cpu and memory
Fig 2: HA support: Cluster or VM Restart HA
© 2020 IBM Corporation 13
HA & DR Approaches : SAP HANA Disaster Recovery
Cold Restart DR solutions (Storage replication)
Site 1 KSYS Site 2
Cold VM
Restart • Slower Recovery
Node 1 Node 1 Node 1 – Additional boot time of VM
Primary DR node DR node
• Cost Savings:
– 2 copies of storage
– No standby LPAR during normal times
Storage Replication
Fig1: VM Restart DR
NO HSR
© 2020 IBM Corporation 14
HA & DR Approaches : SAP HANA HA+DR
HSR
Cold Restart HA+DR solutions
Node 1 or 2 (Storage replication)
Node 1 Node 2 DR node
Primary Secondary
Storage Replication or
Storage Replication
Fig 1: HA & DR : VM Recovery Manager HA + DR
HSR HSR
Node 1 Node 2 Node 1 Node 2
Primary Secondary DR node DR node
Storage Replication
Storage Replication
Fig 2: HA & DR : VM Recovery Manager HA + DR + HA
© 2020 IBM Corporation 15
Demo
© 2020 IBM Corporation 16
Thanks
© 2020 IBM Corporation 17
