Continuous Availability for Business-Critical Data

MetroCluster Technical Overview

Agenda
Causes of Downtime MetroCluster Overview MetroCluster Deployment Scenarios MetroCluster Failure Scenarios New Features Summary

2010 NetApp. All rights reserved.

NetApp Confidential - Limited Use

Customer Challenges
Internal Data Center Failures

External Data Center Failures

FAS HA (Active-Active) MetroCluster storage hardware Addresses the major causes failures within of downtime in the the data data center center

Source: Forester / Disaster Recovery Journal Global Disaster Recovery Preparedness Online Survey, Oct, 2007
2010 NetApp. All rights reserved. NetApp Confidential - Limited Use

Defining Availability and Disaster Recovery

High Availability (HA) refers to eliminating the business impact and downtime due to a hardware failure
RAID-DP Active-Active controllers

Continuous Availability extends HA by providing protection from external events such as power, cooling and network failures, eliminating planned and unplanned downtime and data loss
MetroCluster

Disaster Recovery provides site resiliency protecting against failures that affect an entire data center
SnapMirror
2010 NetApp. All rights reserved. NetApp Confidential - Limited Use

NetApp Data Protection/ Business Continuity Solutions

LAN/WAN Clustering Synchronous Replication Asynchronous Replication Block-Level Incremental Backups Availability Application Recovery
SnapRestore

Data Protection

Continuous Operations Synchronous Clusters MetroCluster Synchronous SnapMirror and Clusters

Synchronous SnapMirror and SyncMirror Asynchronous SnapMirror

Daily Backup

Snapshot Copies Local Restore Remote Restore

Capability

Remote Recovery Low RTO

Cost
5

SnapVault

2010 NetApp. All rights reserved.

NetApp Solution for Hyper-V

Automatic failover within data centers or regional area Practical continuous data availability (CA) and DR for all apps Replication of VMs and data Hyper-V VM and application consistency Asynchronous replication for fast performance Synchronous replication for up-to-the-minute data
NetApp Confidential Internal Use Only 6

2010 NetApp. All rights reserved.

VMware FT and MetroCluster Top-down Continuous availability

2010 NetApp. All rights reserved.

NetApp Confidential - Limited Use

MetroCluster Overview

2010 NetApp. All rights reserved.

NetApp Confidential Internal Use Only

The MetroCluster Difference

Across City or Metro Area Across Floors, Building or Campus Within Data Center

MetroCluster provides continuous availability within a single data center and across data centers in adjacent floors, buildings and metro areas Support for FAS3000, FAS3100, FAS6000 and VSeries

MetroCluster MetroCluster Stretch up to 500 MetroCluster Stretch upm to 100 km

2010 NetApp. All rights reserved.

NetApp Confidential - Limited Use

The MetroCluster Difference

Automatic Protection
Primary Data Center
Exchange

Alternate Data Center

Exchange Oracle

Others

Oracle Sharepoint Home Dirs

Sharepoint Home Dirs

Create Create volume volume

New Volume

New Volume

Oracle

Set Set up up Home Dirs replication replication

Sharepoint New Volume

Create Create Destination Destination Volume Home Dirs Volume Oracle

Sharepoint New Volume

MetroCluster
Exchange

Aggregate Mirroring

Exchange

Create Create volume volume

2010 NetApp. All rights reserved.

Automatic Automatic Mirroring Mirroring

NetApp Confidential Limited Use

The MetroCluster Difference

Quick Recovery time

Primary Data Center
Exchange

1. Break Break Mirror Mirror 1. 2. Bring Bring Online Online 2. 3. Break Break Mirror Mirror 3. 4. Bring Bring Online Online 4. 5. Break Break Mirror Mirror 5. 6. Bring Bring Online Online 6. 7. Break Break Mirror Mirror 7. 8. Bring Bring Online Online 8.

Alternate Data Center

ONLINE ONLINE Oracle ONLINE Home Dirs ONLINE New Volume ONLINE Oracle ONLINE Exchange ONLINE Home Dirs ONLINE Sharepoint Exchange ONLINE Sharepoint

Others

Oracle Sharepoint Home Dirs New Volume

9. Break Break Mirror Mirror 9. 10. Bring Bring Online Online 10.

Oracle

Home Dirs

MetroCluster
Exchange Sharepoint New Volume

New Volume ONLINE

CF CF Forcetakeover Forcetakeover -d -d
2010 NetApp. All rights reserved. NetApp Confidential Limited Use

MetroCluster Deployment Within a Single Data Center, across floors, etc.

Node 1 Stretch MetroCluster Node 2

500m

Volume X

Volume Y

Volume Y Mirror

Data Center

Volume X Mirror

2010 NetApp. All rights reserved.

NetApp Confidential Limited Use

MetroCluster Deployment Across City or Metro Areas

Building A Building B

Fabric MetroCluster <=100 km

DWDM

Dark Fiber

DWDM

Volume X Volume Y Mirror

No SPOFs redundancy in all components and connectivity Dedicated pairs of FC switches

Volume Y Volume X Mirror

2010 NetApp. All rights reserved.

NetApp Confidential - Limited Use

MetroCluster Components
Cluster License Provides automatic failover capability between
sites in case of most hardware failures

SyncMirror Provides an up-to-date copy of data at the

remote site; data is ready for access after failover without administrator intervention

Cluster Remote Provides a mechanism for administrator to

declare site disaster and initiate a site failover via a single command for ease of use

FC Interconnect Provides appliance connectivity between

sites that are greater than 500 meters apart; enables controllers to be located a safe distance away from each other

2010 NetApp. All rights reserved.

NetApp Confidential Limited Use

MC Component: Cluster Remote License

Provides single command failover Provides the ability to have one node take over its partners identity without quorum of disks Root volumes of both appliances MUST be synchronously mirrored By default all volumes are synchronously mirrored and are available during a site disaster Requires administrator intervention as a safety precaution against a split brain scenario (cf forcetakeover d)
IMPORTANT: Site-specific disasters are not the same as a normal cluster failover!

2010 NetApp. All rights reserved.

NetApp Confidential - Limited Use

MC Component: SyncMirror
Writing to an unmirrored volume Writing to a mirrored volume

Two Plexes Created Plex 0 from Pool 0 = Orig Plex 1 from Pool 1 = Mirror

Note: Plex # may change over time

Volume 1 A
Plex 0 Pool 0
2010 NetApp. All rights reserved.

Volume1

A
Plex 0 Pool 0
NetApp Confidential - Limited Use

A
Plex 1 Pool 1

Disk Ownership
Software
Disks assigned to pools by administrator FAS3040, FAS3070, FAS31xx, FAS6000 FAS3020, FAS3050 (Stretch MC only)

Hardware
Determined by physical connection 900 series, FAS3020, FAS3050

2010 NetApp. All rights reserved.

NetApp Confidential - Limited Use

MetroCluster for V-Series

V-Series MetroCluster Up to 30km distance Dedicated or shared SAN Brocade or Cisco Switches LUN limit = Data ONTAP Spindle limit NetApp V-Series Open Storage Controller
Heterogeneous Storage Arrays Refer to V-Series Compatibility Guide at:
http://now.netapp.com/NOW/knowledge/docs/V-Series/supportmatrix/V-Series_SupportMatrix.pdf
2010 NetApp. All rights reserved. NetApp Confidential Limited Use

FAS MetroCluster Up to 100km Dedicated Brocade 504 Max spindles on Fabric Fabric or Stretch Configurations NetApp Storage

Fabric or Stretch Configurations 3rd Party Storage only

MetroCluster Failure Scenarios

2010 NetApp. All rights reserved.

NetApp Confidential Internal Use Only

19

MetroCluster Failure Scenarios: Host Failure

Host 1
Data Center #1 Data Center #2

Host 2

Recovery Automatic
Host #2 will be accessing its data across the data centers

FAS 1

FAS 2

DC1 Primary DC2 Mirror

DC 1 Primary DC 2 Mirror
2010 NetApp. All rights reserved. NetApp Confidential Limited Use

DC2 Primary DC1 Mirror

MetroCluster Failure Scenarios: Storage Controller Failure

Host 1
Data Center #1 Data Center #2

Host 2

Recovery Automatic
FAS 2 provides data services for both data centers

FAS 1

Writes go to both plexes. Reads come from local by default

FAS 2

DC1 Primary DC2 Mirror

DC 1 Primary DC 2 Mirror
2010 NetApp. All rights reserved. NetApp Confidential Limited Use

DC2 Primary DC1 Mirror

MetroCluster Failure Scenarios: Storage Array Failure

Host 1
Data Center #1 Data Center #2

Recovery
Host 2 Single command failover

FAS 1

FAS 2

DC1 Primary DC2 Mirror

DC 1 Primary DC 2 Mirror
2010 NetApp. All rights reserved. NetApp Confidential Limited Use

DC2 Primary DC1 Mirror

MetroCluster Failure Scenarios: Complete Site Failure

Host 1
Data Center #1 Data Center #2

Recovery
Host 2 Single command failover

FAS 1

FAS 2

DC1 Primary DC2 Mirror

DC 1 Primary DC 2 Mirror
2010 NetApp. All rights reserved. NetApp Confidential Limited Use

DC2 Primary DC1 MIrror

MetroCluster Failure Scenarios: Interconnect Failure

Host 1
Data Center #1 Data Center #2

Recovery
Host 2
No failover; mirroring disabled Both appliance heads will continue to run serving their LUNs/volumes Resyncing happens automatically after interconnect is reestablished

FAS 1

FAS 2

DC1 Primary DC2 Mirror

DC 1 Primary DC 2 Mirror
2010 NetApp. All rights reserved. NetApp Confidential Limited Use

DC2 Primary DC1 Mirror

MetroCluster Failure Scenarios: Rolling Disasters

Host 1
Data Center #1 Data Center #2

Recovery
Host 2 Single command failover

Loss of Entire Data Center

FAS 1

Loss of Network

FAS 2

DC1 Primary DC2 Mirror

DC 1 Primary DC 2 Mirror
2010 NetApp. All rights reserved. NetApp Confidential Limited Use

DC2 Primary DC1 Primary

MetroCluster Failure Scenarios: Return to normal

Host 1
Data Center #1 Data Center #2

Host 2

One-step return to normal CF giveback

FAS 1

Mirror is rebuilt DC1 primary reverts back to DC1

FAS 2

DC1 Primary DC2 Mirror

DC 1 Primary DC 2 Mirror
2010 NetApp. All rights reserved. NetApp Confidential Limited Use

DC2 Primary DC1 Primary Mirror

Support Matrix
Model 3020 3040 3050 3070 3140 3170 6030 6040 6070 6080 Stretch Supported Max. Disks YES 168 YES 336 YES 336 YES 504 YES 420 YES 840 YES 840 YES 840 YES 1008 YES 1166 Fabric Supported Max. Disks YES 168 YES 336 YES 336 YES 504* YES 420 YES 672* YES 672* YES 672* YES 672* YES 672*

* 504 spindle count requires Data ONTAP 7.2.4 or later * 672 spindle count requires Data ONTAP 7.3.2 or later
2010 NetApp. All rights reserved. NetApp Confidential - Limited Use

31xx Configurations

NOTE: 1.Single controller chassis only 2.FC/VI card required in both Stretch and Fabric configuration
2010 NetApp. All rights reserved. NetApp Confidential - Limited Use

Deduplication for MetroCluster

Deduplication provides cost savings for MetroCluster
Deduplication efficiencies for both the source and mirrored arrays Investment protection and storage efficiency for other vendor storage with V-Series

PVR no longer required

Must be licensed on both nodes Must be running Data ONTAP 7.2.5.1 or later, 7.3.1 or later Subject to same considerations as non-MetroCluster systems

2010 NetApp. All rights reserved.

NetApp Confidential - Limited Use

Traffic Isolation: Available since 7.2.6

Traffic isolation w/dual untrunked ISLs

2010 NetApp. All rights reserved.

NetApp Confidential - Limited Use

Best Practices: MetroCluster

Choose the right controller Connections are important
Calculate distances properly Use the right fiber for the job Verify correct SFPs, cables required, and patch panels

Remember mirroring requires 2x disks Remember MetroCluster spindle maximums Remember speed restrictions

2010 NetApp. All rights reserved.

NetApp Confidential - Limited Use

MetroCluster Complements Host Clustering

Solution Highlights
No SPOF High failure resiliencyno failover required for any single component failures: Disks, disk shelves, disk loop, MetroCluster switches, ISLs, cables, interconnects, single Exchange node, appliance head No storage failover required, even in case of failure of both Exchange nodes No/minimal disruption to service availability for any of above failure conditions* Failover to the DR site within minutes after sitewide failure No loss of data Easy failover operation aided by server cluster and storage cluster Seamless backup and recovery across cluster resources with SME Support of active-active configurationsfully utilized HW at both sites Minimal performance degradation during and after recovery

MetroCluster works with other host clustering software such as Microsoft Cluster Server, Veritas Cluster Server, and SunCluster.
2010 NetApp. All rights reserved. NetApp Confidential - Limited Use

Recent Additions (7.3.2)

Up to 672 spindles on Fabric MC Multiple Subnet Support Nondisruptive disk shelf replacement Nondisruptive upgrades

2010 NetApp. All rights reserved.

NetApp Confidential - Limited Use

Multiple Subnet Support

2010 NetApp. All rights reserved.

Whats New?
MetroCluster TieBreaker Solution Vmware Certification with HA Vmware Certification with FT

2010 NetApp. All rights reserved.

NetApp Confidential - Limited Use

MetroCluster Tiebreaker Solution

Automates site failover (cfod) Developed by Professional Services and RRE Supported via RRE Available via PVR only Not required (or necessary) in most instances Requirements
Operations Manager Perl installed on Ops Mgr Server RLM (optional for fencing)
2010 NetApp. All rights reserved.

TieBreaker Diagram

2010 NetApp. All rights reserved.

NetApp Confidential Limited Use

MetroCluster for VMware Environments

Delivers continuous data availability
ESX Server ESX Server

ESX Server ESX Server ESX Server

Complements VMware host environment with complete server to storage availability

VMotion Technology

Improves read performance

Interleave reads from mirrored arrays

Simplifies operations
Zero interdependencies

Deploys with confidence

MetroCluster

Tested interoperability

Leverages Virtualization Guarantee Program

2010 NetApp. All rights reserved. NetApp Confidential - Limited Use

MetroCluster Reference Stories

Lee County Court of Clerks Public Sector, N.A
Full Story: http://media.netapp.com/documents/LeeCounty.pdf

United Health Network Healthcare, N.A

Full Story: http://media.netapp.com/documents/UHN.pdf

Bitburger Brewery Group Retail, Germany

Full story: http://www.netapp.com/us/library/customer-stories/bitburger.html

New customer stories coming soon

2010 NetApp. All rights reserved.

NetApp Confidential - Limited Use

39

Challenges
Business requiring 24x7 operations Risk of downtime was unacceptable Replace its ERP system based on Oracle with an ERP infrastructure based on SAP Implement a VMware server virtualization solution

Solution
FAS6080 and FAS3140 storage systems Located at opposite ends of the data center and configured as a stretch MetroCluster Provides Lexmark with a local campus disaster recovery and high availability solution Deduplicate data from Oracle Initial tests have shown a 60% space savings
NetApp Confidential - Limited Use

Stretch MetroCluster
Data Center

2010 NetApp. All rights reserved.

Proposing a MetroCluster
Information to be gathered
Distance between locations Number of disks consider mirroring, growth

New system or upgrade? Upgrade considerations

Hardware or software ownership Existing data 31xx issues

2010 NetApp. All rights reserved.

NetApp Confidential Limited Use

MetroCluster Positioning
Qualification questions
Is continuous data availability important to one or more of your product applications? Would providing geographic protection be of value? What distance between controllers would provide the best protection (building, floor campus, metro, or regional)? If you are employing an HA solution for a virtualized environment, why not have the same resilience at the storage level?
2010 NetApp. All rights reserved. NetApp Confidential Limited Use

SnapMirror Sync versus MetroCluster

Feature
Replication Network Limit on concurrent transfers Distance limitation Replication between HA pairs Manageability Multi-use replica

SnapMirror Sync
IP or FC Yes Up to 200 KM (latency driven) Yes FilerView, CLI, Ops Mgr Yes (FlexClone) No (dump) Dedupe volume and sync volume cannot be in the same aggregate Yes

MetroCluster
FC only No limit 100 KM No CLI Limited (app can read from both plexes)

Deduplication

No restrictions

Use of secondary node for async mirroring

2010 NetApp. All rights reserved.

No. Async replication has to occur from primary plex

Best Practices: MetroCluster

Choose the right controller
Based on normal sizing tools and methods

Connections are important

Calculate distances properly Use the right fiber for the job Verify correct SFPs, cables required, and patch panels

Remember mirroring requires 2x disks Remember MetroCluster spindle maximums Remember speed restrictions
2010 NetApp. All rights reserved. NetApp Confidential Limited Use

References
TR3548 - MetroCluster Design and Implementation Guide
http://media.netapp.com/documents/tr3548.pdf

TR3517 MetroCluster Upgrade Planning Guide

http://media.netapp.com/documents/tr-3517.pdf

TR3552 Optical Network Installation Guide

http://media.netapp.com/documents/tr-3552.pdf

TR3606 - High Availability and Disaster Recovery for VMware using NetApp SnapMirror and MetroCluster
http://media.netapp.com/documents/tr-3606.pdf

Data ONTAP Active/Active Configuration Guide

http://Now.netapp.com

MetroCluster Compatibility Matrix

http://now.netapp.com/NOW/knowledge/docs/olio/guides/metrocl uster_compatibility/

2010 NetApp. All rights reserved.

NetApp Confidential - Limited Use

45

Value of MetroCluster
Simple
Set it Once: Simple to deploy, with no scripts or dependencies on the application or operating system Zero change management through automatic mirroring of changes to user, configuration, and application data
Data Center

Continuous Availability
Zero unplanned downtime through transparent failover with protection for hardware plus network and environmental faults Zero planned downtime through nondisruptive upgrades for storage hardware and software

Cost Effective
50% lower cost, no host-based clustering required, added efficiency through data deduplication and server virtualization
2010 NetApp. All rights reserved. NetApp Confidential - Limited Use

MetroCluster

Thank You

Copyright 2009 NetApp, Inc. All rights reserved. No portions of this document may be reproduced without prior written consent of NetApp, Inc. NetApp, the NetApp logo, Go further, faster, Data ONTAP, NOW, RAID-DP, SnapMirror, and SyncMirror are trademarks or registered trademarks of NetApp, Inc. in the United States and/or other countries. Microsoft is a registered trademark of Microsoft Corporation. Oracle is a registered trademark of Oracle Corporation. Veritas is a trademark of Symantec Corporation or its affiliates in the U.S. and other countries. SAP is a registered trademark of SAP AG. VMware and VMotion are registered trademarks of VMware, Inc. All other brands or products are trademarks or registered trademarks of their respective holders and should be treated as such.