Virtualize Your Disaster! Remove the Disaster from “Disaster Recovery”

Agenda

Challenges of Traditional Disaster Recovery

Improving Disaster Recovery with VMware Virtualization

Impact of VMware Virtualization on Disaster Recovery

Defining Disaster Recovery

What is a “disaster”?

Extended outage that requires an organization to recover IT services at an alternate or rebuilt site

What is not a disaster?

Failure of an individual server

A short service interruption

Power failure Server failure Disk failure Fire Hurricane Storage failure Flood Network failure Earthquake

Requirements for Disaster Recovery Minimize Downtime 93% of companies that lost their data center for ten days or more due to a disaster filed for bankruptcy within one year of the disaster. --National Archives and Records Administration Control Cost Minimize Risk 73% of executives expressed concern with the costs associated with maintaining a secondary data centre --Beacon Technology Partners 92% of users surveyed acknowledged that their companies would face serious consequences if they had to implement their disaster recovery plans. --Dynamic Markets Ltd. Effective disaster recovery is a business imperative, but is very difficult to achieve

Current State of Physical Disaster Recovery

Disaster recovery protection tiered according to business priorities

Recovery requirements continue to become more demanding

Tier RPO RTO Cost I Immediate Immediate $$$ II 24+ hrs. 48+ hrs. $$ III 7+ days 5+ days $

Physical Server Protection

Multiple elements to protect to ensure recoverability

Requires multiple technologies and processes

System disk backups

Backups and/or replication for applications and data

Tracking of hardware configuration requirements

Storage Server Operating System Application Application installation Application configuration OS configuration OS customizations Hardware configuration Firmware level & settings System disk Application data

Disaster Recovery for Physical Infrastructure

Fastest, easiest recovery requires duplicate datacenter

Servers

Rack space

Network infrastructure

Storage infrastructure

Power and cooling infrastructure

Duplicating datacenter creates ongoing burdens

Ongoing management to keep identical to production

Idle except for occasional tests

Production Recovery

Disaster Recovery Process for Physical Infrastructure Production Recovery

Copy important data

Hardware configuration

System disk

Application installation

Application data

Transfer data to recovery site

Backup tapes

CD’s/DVD’s

Images (e.g. Ghost)

Replication

Start system and application recovery

OS recovery

Configuration

Data recovery

Testing

WAN

Complications of Physical Recovery

Variety of information and data to protect

Separate processes for system and data disks

No easy way to track server configuration

Tapes can be misplaced, mislabeled, or corrupted

Complex recovery process

Bare metal recovery procedures

Failed recoveries due to hardware dependencies

Many manual steps prone to user errors

Inability to test recovery sufficiently

Additional infrastructure

Disruption to operations

Recovery

Challenges of Disaster Recovery

Minimize downtime

Reduce risk

Control cost

Many manual processes for recovery

Multiple steps to overcome hardware differences

Incomplete or out-of-date runbooks

Testing requires additional hardware and infrastructure

Usually only data is regularly and cleanly updated

Frequent failures during recovery

Simplest recovery requires identical hardware

Idle recovery hardware is impossible to repurpose

Multiple third-party products necessary for recovery

Agenda

Properties of virtualization

Data protection with virtualization

Disaster recovery automation

VMware Vision for Disaster Recovery Rapid

Automate recovery process

Eliminate failures due to hardware dependencies

Integrate different components of recovery

Reliable

Enable easier, more frequent testing

Turn manual, inconsistent processes into pre-programmed, repeatable processes

Manageable

Centralize and simplify management of recovery plans

Make disaster recovery protection a property of virtual infrastructure

Affordable

Eliminate idle recovery hardware

Eliminate dependencies on physical infrastructure

OS in virtual machine sees “virtual hardware” (CPU, SCSI disk, memory, network)

Virtualization layer maps virtual hardware to physical hardware

Virtual hardware remains identical regardless of physical hardware

Virtual machines are hardware independent and can run on any server without modification Physical Hardware Virtual Machine Virtualization Layer OS Virtual Machine OS

Everything about a system stored in small number of files

Simplifies copying and cloning of systems

Simplifies provisioning systems for recovery and recovery testing

Virtual machines encapsulate and entire system in a few files on disk System Apps = files on disk Data Virtual Machine Physical Server

Virtualized Server Protection

Everything needed to recover a virtual machine is stored in a few files in a single folder on disk

Hardware configuration (.vmx)

Firmware settings (.nvram)

Operating system install (.vmdk)

Application install (.vmdk)

Application data (.vmdk)

Standard backup and replication tools can be used to protect virtual machines

Virtual Machines VMFS Virtual Machine Servers ESX ESX ESX ESX

Data Protection Options for Virtual Machines

Virtual machine backup

File and image backup

Useful for non-production and many production virtual machines

Slowest recovery time

Server-based replication

Granular replication

Often OS-dependent

Processing overhead on server

Array-based replication

Minimal overhead on server

LUN-level granularity

Storage Servers Virtual Machines ESX ESX ESX Server-based replication Array-based replication Virtual machine backups

Backup for Virtual Machines

Hardware-independent backup and restore

No bare metal recovery or operating system reinstall required

Simplified backup of system state, boot disk, and data disks

No need for separate tools and procedures

Non-disruptive backup via Consolidated Backup interface

Image and file-level backup

Minimizes backup load on ESX and network

Integrated with leading 3 rd -party backup products

Backup Mount Snap Snap Storage ESX server Backup Server

Virtualized Recovery Infrastructure

Servers running VMware Infrastructure as recovery targets

Can recover a VM to any VMware Infrastructure edition

Virtual machine image

VMware Converter to create virtual machine images of physical systems

Backup or replication software to create copies of virtual machines

Virtual Machines ESX ESX Recovery Production Converter ESX

VMware Virtualization Simplifies Infrastructure Requirements

Eliminate need for identical infrastructure for recovery

VMware Infrastructure provides hardware-independent recovery

Reduces ongoing management burden of recovery site

Reduce infrastructure requirements

Consolidate production and recovery servers

Waterfall servers from production to recovery

VMware Production VMware Recovery

Simplifying the Disaster Recovery Process

Eliminate recovery steps

No operating system re-install or bare-metal recovery

No time spent reconfiguring hardware

Standardize recovery process

Consistent process independent of operating system and hardware

Configure hardware Install OS Configure OS Install backup agent Start “Single-step automatic recovery” Restore VM Physical Virtual 40+ hrs < 4 hrs Power on VM

Simpler Disaster Recovery Testing with Virtualization

Snapshot and clone replicated data to create testing VM’s

Connect test VM’s to an isolated network

Power up testing VM’s to validate recovery

Delete VM clones used for testing

SAN Recovery Site Data. OS Application Powered On DR VMs Live DR Test DR Replicated Data Snapshot Target VMFS OS.image Appln.image Data.image 15 GHz 9 GHz

Simplifies and automates disaster recovery workflows:

Setup, testing, failover

Turns manual recovery runbooks into automated recovery plans

Provides central management of recovery plans from VirtualCenter

Automating the Disaster Recovery Process

Works with VMware Infrastructure to make disaster recovery rapid, reliable, manageable, affordable

Site Recovery Manager provides management and automation of the disaster recovery process for virtual machines

Site Recovery Manager Key Components Storage Servers VMware Infrastructure Virtual Machines VirtualCenter Site Recovery Manager Storage Servers VMware Infrastructure Virtual Machines VirtualCenter Site Recovery Manager

Site Recovery Manager

Manages and monitors recovery plans

Tightly integrated with VirtualCenter

Storage Partner Replication

Integrated via replication adapters created, certified and supported by replication vendor

Partner Replication

VMware Infrastructure

Requires ESX server 3.0.2, ESX 3.5 U1 or later

Requires VirtualCenter 2.5 U1 or later

Storage

iSCSI or FibreChannel storage

Disaster Recovery Setup

Integrate with replication

Identify which virtual machines are protected by replication configuration

Map recovery resources

Server resources, network resources, management objects

Create recovery plans

For virtual machines, applications, business units

Convert manual runbook to pre-programmed response

Customizable with scripting and callouts

Testing

Replication Management

Snapshot replicated LUNs before test

Delete snapshots of replicated LUNs after test

Network Management

Change all virtual machines to a test port group before powering them on

Customization/extensibility

Same breakpoints and callouts as failover sequence

Extra breakpoints and callouts around the test bubble

Failover Automation

Detect site failures

Raise alert when heartbeat lost

Initiate failover

User confirmation of outage

Granular failover initiation

Manage replication failover

Break replication

Make replica visible to recovery hosts

Execute recovery process

Use pre-programmed plan

Provide visibility into progress

Agenda

Hancock Bank: Before Virtualization Production Datacenter DB HR App App DR Hosting Provider Site DB HR App App Backup data

Unknown hardware makes and models

Recovery time unable to meet requirements

Failing back to primary site takes months

Hancock Bank: With Virtualization Production Datacenter DR Hosting Provider DB HR App App

Recovery to any hardware

Recovery time within 24 hour objective

Failback takes days, not months

Image and data backups DB HR App App “ Without VMware Infrastructure, it would have taken us weeks to recover our critical systems when Hurricane Katrina hit our datacenter. VMware Infrastructure enabled us to get our critical systems up and running within 24 hours.” -- Scott Fontenette, Hancock Bank

VMware Impact on Disaster Recovery