High Availability and Disaster Recovery using Virtualisation

1
High Availability and Disaster Recovery using
Virtualisation
info_at_techarchsolutions.com www.techarchsolutions.c
om
2
Definitions

• HA ability to proactively avoid some types of
  disasters before they occur
• DR ability to quickly recover data and restart
  business operations in the event of a disaster

3
Traditional HA solutions - Clustering

• Purpose
• High Availability
• Scalability/Load Balancing
• Manageability
• Types
• Active/Active
• Active/Passive
• Cold Failover
• Hot Failover

4
Traditional HA solutions - Clustering

• Benefits
• Reduced probability of disaster occurrence
• Reduced planned downtime windows and costs
• Improved operational efficiencies
• Disadvantages
• Expensive IT investment
• Active / Passive clusters waste hardware
• Complex implementation

5
Current DR solutions

• Media-based backup/restore
• Good Familiar, cheap
• Bad slow, high cost of downtime
• Cold Standby
• Good server availability
• Bad wasted hardware
• Hot Standby
• Good quick turnaround time, multitasking
• Bad expensive

6
What is Virtualisation?

• Multiple virtual systems residing on a physical
  server.
• Provides a layer of abstraction between the
  hardware and software that runs on it.

7
Virtualisation as an Option

• Disadvantages
• Not recommended for servers that are heavily
  utilised
• Not recommended for applications/servers
  requiring specific hardware requirements
• If designed badly, single point of failure
• Advantages
• Greater solution ROI
• Savings on hardware and rack space
• Ease of management
• Efficiency in deployment of VMs
• Zero-downtime hardware maintenance
• Higher SLA and DR at a lower cost

8
Clustering for Higher Availability in a Virtual
Infrastructure

• Cluster Virtual Machines for High Availability
• Cluster host servers for High Availability
• Application Clustering
• Virtual Infrastructure can be a replacement for
  clusters where clustering for H/W failure only

9
Virtual Machine HA solutions

• VMs can be run from any server in a pool
• VMs can be dynamically moved between servers

10
Virtual Machine HA solutions

• VMs from a failed server can be brought back
  on-line in remaining server pool quickly
  (within/across DCs)

11
Clustering with Virtual Machines

• Cluster between physical and virtual machines
• Cluster-in-a-box
• VM Clustering between 2 Host Servers

12
Physical-to-Virtual Clustering

• Physical machine clustered with a Virtual Machine
• Shared storage required

13
Cluster in a Box

• One VM is primary and second is backup
• Quorum on shared storage

14
VM clustering between two servers

• VM on one host is primary, VM on second host is
  backup
• Shared storage required

15
Virtual DR solutions

• VMotion between DCs
• SAN mirroring between DCs
• Snapshot and copy between DCs
• Snapshot to tape or file server
• Physical to virtual DR

16
Virtualisation Vendors

• Xen
• Solaris Zoning
• Microsoft Virtual Server 2005
• VMware ESX

17
Migration Tools for Virtualisation

• Microsoft VSMT
• Leostream PgtV direct
• VMware P2V Assistant
• Platespin PowerConvert

18
PlateSpin PowerConvert
19
PlateSpin PowerConvert Solutions

• Consolidation
• Recovery
• Test lab automation
• Business Service Provisioning
• Hardware migrations
• Data Centre Moves or Centralisation

20
Future of Virtual Infrastructure

• Host server clustering ability to automatically
  bring up VMs on secondary server in the event of
  failure
• Integration of the system failure detection card
  into the ESX layer
• Self managing Data Centres
• Application virtualisation (Softgrid Softricity)

21
QUESTIONS

• ???????????????????????????