1. Continuous Availability for Business-Critical Data MetroCluster Technical Overview

2. NetApp Confidential - Limited Use
Agenda
Causes of Downtime
MetroCluster Overview
MetroCluster Deployment Scenarios
MetroCluster Failure Scenarios
New Features
Summary
NetApp Confidential - Limited Use

3. Customer Challenges Internal Data Center Failures External Data Center
MetroCluster
Addresses the major causes
of downtime in the data center
FAS HA
(Active-Active)
Addresses storage hardware failures within the data center
Let’s first understand the customer challenges.
Challenging economic times are compelling businesses to achieve even greater levels of cost savings and operational efficiency. Yet business-critical applications still require vital data to be protected and available to meet increasing service-level demands. The majority of businesses that fail to protect their critical applications do not get a second chance, and those that fail to reduce their operational expenses may suffer the same fate.
This chart represents the results of an online survey conducted by Forester Research and Disaster Recovery Journal that included responses from 250 IT decision makers.
The key takeaways:
The overwhelming majority of declared disasters are caused by internal data center failures. In fact, more than 75% of those surveyed have experienced an outage inside the data center. These common causes of downtime include things such as power, hardware, and network. While, the more catastrophic disasters such as flood, hurricane, and fire that affect the entire data center are rare occurrences.
It is important to note that the NetApp FAS HA solution which includes the Active-Active Controller Configurations only address storage hardware failures. (click for animation) Only NetApp MetroCluster can protect the data against these major causes of downtime inside the data center.
Let’s take a closer look at MetroCluster…
Source: Forester / Disaster Recovery Journal Global Disaster Recovery Preparedness Online Survey, Oct, 2007
NetApp Confidential - Limited Use 3

4. 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®
NetApp Confidential - Limited Use

5. NetApp Data Protection/ Business Continuity Solutions
Continuous Operations
Availability
LAN/WAN Clustering
Synchronous Replication
Synchronous
Clusters
MetroCluster
Asynchronous Replication
Synchronous
SnapMirror® and Clusters
Block-Level Incremental Backups
Synchronous SnapMirror and SyncMirror®
Asynchronous SnapMirror
Cost
NetApp Software Architecture
The purpose of this slide is to demonstrate that NetApp has Data Protection/Business continuity solutions for all levels of availability requirements.
SnapVault®
Application Recovery
SnapRestore®
Daily
Backup
Snapshot™
Copies
Local Restore
Remote Restore
Remote Recovery – Low RTO
Capability 5

6. NetApp Solution for Hyper-V
MetroCluster for Continuous Data Availability
MetroCluster can address your Hyper-V customer’s continuous data availability requirements, whether it is deployed inside a data center; at different locations in a building or across metro-wide deployments. Automatic failover of storage and VM data between Data Centers A and B – which can be up to 100km apart – complements Hyper-V’s server and application failover capabilities. MetroCluster seamlessly maintains application data availability in the face of a total storage outage due to losses of power, cooling, or network connectivity; storage array shutdown; or operational error.
As an array-based active-active clustered solution, MetroCluster eliminates the need for complex failover scripts, server reboots, or application restarts. This enables a level of availability that goes beyond the high-availability features in a single array. MetroCluster supports NetApp FAS3000, FAS3100, FAS6000 and V-Series.
SnapMirror for Cost Effective Disaster Recovery
Hyper-V customers can take data protection a step further. They can protect themselves from a site-wide outages by deploying SnapMirror for VM data replication across unlimited distances. SnapMirror mirrors data asynchronously at high speeds over WAN connections, protecting critical Microsoft applications from regional disruptions. SnapMIrror also supports synchronous replication for customers who cannot tolerate any data loss.
. Unlike other solutions, SnapMirror is flexible enough and simple enough to use across all applications in the environment, not just on the most critical applications. SnapMIrror also reduces DR costs by enabling replication between FC and SATA systems.
Application consistency is also maintained by leveraging application-aware Snapshot copies, as discussed on the previous slide. SnapMIrror also uses SnapRestore technology for recovery of entire virtual Microsoft infrastructures in minutes.
For a comprehensive business continuity solution for virtual Microsoft environments, customers can achieve continuous availability protection against data center failures with MetroCluster and Disaster Recovery protection against the loss of an entire data center with SnapMIrror.
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

7. VMware FT and MetroCluster Top-down Continuous availability
NetApp Confidential - Limited Use

8. MetroCluster Overview
NetApp Confidential – Internal Use Only

9. The MetroCluster Difference
Across City or Metro Area
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 V-Series
Across Floors, Building or Campus
Within Data Center
MetroCluster – Stretch up to 100 km
MetroCluster can address your continuous availability requirements, whether it is deployed inside a data center; (click for animation) at different locations in a building; (click for animation) or across city or metro-wide deployments up to a maximum distance of 100km. This enables a level of availability that goes beyond the high-availability features in a single array which makes MetroCluster a highly versatile solution.
MetroCluster supports NetApp FAS3000, FAS3100, FAS6000 and V-Series.
MetroCluster
MetroCluster – Stretch up to 500 m
NetApp Confidential - Limited Use

10. The MetroCluster Difference Automatic Protection
Primary Data Center
Alternate Data Center
Oracle
Oracle
Exchange
Exchange
Others
Sharepoint
Sharepoint
Home Dirs
New Volume
New Volume
Create volume
Set up replication
Create Destination Volume
Oracle
Exchange
Home Dirs
Sharepoint
Oracle
Exchange
Home Dirs
Sharepoint
MetroCluster
New Volume
Aggregate Mirroring
Create volume
Automatic Mirroring
NetApp Confidential – Limited Use

11. The MetroCluster Difference Quick Recovery time
Primary Data Center
Alternate Data Center
1. Break Mirror
2. Bring Online
ONLINE
Exchange
Exchange
Oracle
Oracle
3. Break Mirror
4. Bring Online
ONLINE
Others
Sharepoint
Sharepoint
ONLINE
5. Break Mirror
6. Bring Online
Home Dirs
Home Dirs
7. Break Mirror
8. Bring Online
ONLINE
New Volume
New Volume
9. Break Mirror
10. Bring Online
ONLINE
Oracle
Exchange
Home Dirs
Sharepoint
CF Forcetakeover -d
ONLINE
Oracle
Exchange
Home Dirs
Sharepoint
MetroCluster
New Volume
NetApp Confidential – Limited Use

12. NetApp Confidential – Limited Use
MetroCluster Deployment Within a Single Data Center, across floors, etc.
Node 1
Node 2
Stretch MetroCluster
500m
Cluster heartbeat is through InfiniBand
Uses the NVRAM cards EXCEPT
X1926A card for 31xx series
<= 500 m when using point-to-point connection*
Fibre Channel cabling
Patch panels reduce the distance
Disk shelves
Supports up to platform limit
2Gb or 4Gb
ATA supported
Disk ownership same as Fabric MetroCluster
*Must use OM3 type cabling (or better) to achieve 500 m
Volume X
Volume Y
Volume Y Mirror
Volume X Mirror
Data Center
NetApp Confidential – Limited Use

13. MetroCluster Deployment Across City or Metro Areas
Building A
Building B
Fabric MetroCluster
Volume X
Volume Y Mirror
<=100 km
Volume Y
Volume X Mirror
DWDM
Dark Fiber
No SPOFs – redundancy in all components and connectivity
Dedicated pairs of FC switches
Cluster Interconnect is VI over FC (versus SCSI)
X1922A card for 3020, 3050
X1926A card for 3040, 3070, 31xx, and 6xxx
Fibre Channel Switches
Disk and controller interconnect
Currently using Brocade 300 and 5100, Licensed for full (multiswitch) fabric
Customer-supplied switches not supported
Configuring for Long Distances
Up to 10 km - 4 “Long wave” SFPs (X1670A-R5)
Greater than 10 km - 4 “Extended long wave” SFPs; Brocade certified
Extended distance license required in order to set Buffer credits set accordingly
Storage ports can be
2Gbps - 2 dual-port FC HBAs OR 4 onboard ports
4Gbps - 4 onboard ports (3070, 6000 series)
Quad-port FC HBAs
Disk shelves
Shelves on each loop must be same speed
2 shelves per loop maximum
ATA shelves not supported
Depending on FAS model, ownership determined by software or hardware rules
Are attached to same ports on both switches (hardware ownership)
NetApp Confidential - Limited Use

14. 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
NetApp Confidential – Limited Use

15. MC Component: Cluster Remote License
Provides single command failover
Provides the ability to have one node take over its partner’s 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!
Cluster_remote is the software based license to enable the cf takeover –d command.
This is the “metro-cluster license”
NetApp Confidential - Limited Use

16. 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
Volume1
Volume1
Pool 0
Plex 0 A
Pool 0
Plex 0 A
Pool 1
Plex 1 A’
NetApp Confidential - Limited Use

17. NetApp Confidential - Limited Use
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
NetApp Confidential - Limited Use

18. MetroCluster for V-Series
V-Series MetroCluster
FAS MetroCluster
Up to 30km distance
Up to 100km
Dedicated or shared SAN
Dedicated
Brocade or Cisco Switches
Brocade
LUN limit = Data ONTAP Spindle limit
504 Max spindles on Fabric
Fabric or Stretch Configurations
3rd Party Storage only
NetApp Storage
There are times when your customers want to protect their existing storage investment. And with NetApp’s V-Series storage virtualization appliance, your customers can deploy a heterogeneous continuous availability solution with MetroCluster for their entire storage infrastructure.
Use V-Series with deduplication as a door-opener for storage efficiency
You can also use the Virtualization Guarantee Program to close the deal.
NetApp
V-Series
Open Storage
Controller
Heterogeneous Storage Arrays
Refer to V-Series Compatibility Guide at:
NetApp Confidential – Limited Use

19. MetroCluster Failure Scenarios
NetApp Confidential – Internal Use Only

20. MetroCluster Failure Scenarios: Host Failure
Recovery – Automatic
Host #2 will be accessing its data across the data centers
Data Center #1
Data Center #2
Host 1
Host 2
FAS 1
1. Assuming host/application clustering the application fails over to host #2. This is the same as if it was standard filer cluster with the exception that host #2 will be accessing it’s data across the data centers.
FAS 2
DC1 Primary
DC2 Primary
DC 1 Primary
DC2 Mirror
DC1 Mirror
DC 2 Mirror
NetApp Confidential – Limited Use

21. MetroCluster Failure Scenarios: Storage Controller Failure
Recovery – Automatic
FAS 2 provides data services for both data centers
Data Center #1
Data Center #2
Host 1
Host 2
Writes go to both plexes.
Reads come from local by default
FAS 1
The same as a normal CFO because filer #2 can access the disks in both DCs to successfully complete the take-over.
FAS 2
DC1 Primary
DC2 Primary
DC 1 Primary
DC2 Mirror
DC1 Mirror
DC 2 Mirror
NetApp Confidential – Limited Use

22. MetroCluster Failure Scenarios: Storage Array Failure
Recovery
Single command failover
Data Center #1
Data Center #2
Host 1
Host 2
FAS 1
1. No automated filer takeover occurs. Filer #2 will continue to run serving it’s LUNs/volumes.
2. Application may or may not failover to Host #2 depending on clustering software being used. NO access to data being server by filer#1.
3. Manual cfo –d command causes filer-cluster takeover to happen.
Cluster code will ignore quorum issues (due to ½ of the partner mailbox disks missing).
FAS 2
DC1 Primary
DC2 Primary
DC 1 Primary
DC2 Mirror
DC1 Mirror
DC 2 Mirror
NetApp Confidential – Limited Use

23. MetroCluster Failure Scenarios: Complete Site Failure
Recovery
Single command failover
Data Center #1
Data Center #2
Host 1
Host 2
FAS 1
1. No automated filer takeover occurs. Filer #2 will continue to run serving it’s LUNs/volumes.
2. Application may or may not failover to Host #2 depending on clustering software being used. NO access to data being server by filer#1.
3. Manual cfo –d command causes filer-cluster takeover to happen.
FAS 2
DC1 Primary
DC2 Primary
DC 1 Primary
DC2 Mirror
DC1 MIrror
DC 2 Mirror
NetApp Confidential – Limited Use

24. MetroCluster Failure Scenarios: Interconnect Failure
Recovery
No failover; mirroring disabled
Both appliance heads will continue to run serving their LUNs/volumes
Resyncing happens automatically after interconnect is reestablished
Data Center #1
Data Center #2
Host 1
Host 2
FAS 1
1. No automated filer takeover occurs. Both filer heads will continue to run serving it’s LUNs/volumes.
2. Filer Status? Both filer heads are still running. Not a problem because each head is “only” serving it’s LUNs/volumes
3. Assuming that each host is accessing it’s data locally (i.e. host #1 accessing LUNs served by filer #1, host #2 accessing LUNs served by filer #2) applications continue to run without problems.
4. Re-syncing happens automatically after interconnect is reestablished.
5. Host cluster will probably failover since they cannot communicate
FAS 2
DC1 Primary
DC2 Primary
DC 1 Primary
DC2 Mirror
DC1 Mirror
DC 2 Mirror
NetApp Confidential – Limited Use

25. MetroCluster Failure Scenarios: Rolling Disasters
Loss of Entire Data Center
Recovery
Single command failover
Data Center #1
Data Center #2
Host 1
Host 2
Loss of Network
FAS 1
1. No automated filer takeover occurs. Filer #2 will continue to run serving it’s LUNs/volumes.
2. Application may or may not failover to Host #2 depending on clustering software being used. NO access to data being server by filer#1.
3. Manual cfo –d command causes filer-cluster takeover to happen.
FAS 2
DC1 Primary
DC2 Primary
DC 1 Primary
DC2 Mirror
DC1 Primary
DC 2 Mirror
NetApp Confidential – Limited Use

26. MetroCluster Failure Scenarios: Return to normal
One-step return to normal
CF giveback
Data Center #1
Data Center #2
Host 1
Host 2
Mirror is rebuilt
DC1 primary reverts back to DC1
FAS 1
FAS 2
DC1 Primary
DC2 Primary
DC 1 Primary
DC2 Mirror
DC1 Mirror
DC1 Primary
DC 2 Mirror
NetApp Confidential – Limited Use

27. Support Matrix Stretch Fabric Model Supported Max. Disks 3020 YES 168
3040
336
3050
3070
504
504*
3140
420
3170
840
672*
6030
6040
6070
1008
6080
1166
* 504 spindle count requires Data ONTAP® or later
* 672 spindle count requires Data ONTAP or later
NetApp Confidential - Limited Use

28. NetApp Confidential - Limited Use
31xx Configurations
NOTE:
Single controller chassis only
FC/VI card required in both Stretch and Fabric configuration
NetApp Confidential - Limited Use

29. 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® or later, or later
Subject to same considerations as non-MetroCluster systems
NetApp deduplication offers storage efficiency across all storage tiers: primary data, backup data, and archival data. NetApp deduplication is included with all NetApp FAS3000, FAS3100, FAS6000, and V-Series systems. Unlike other deduplication solutions, no additional hardware or software licenses are required, and NetApp deduplication is application and protocol agnostic.
Based on customer feedback and AutoSupport data, customers are, on average, realizing 25 percent to 30 percent capacity savings. And 50 percent to 90 percent savings in virtualized environments. MetroCluster practically pays for itself with the savings customers achieve with deduplication in a virtualized environment.
NetApp Confidential - Limited Use

30. Traffic Isolation: Available since 7.2.6
Traffic isolation w/dual untrunked ISLs
NetApp Confidential - Limited Use

31. 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
Controller sizing for MetroCluster is the same as a standard active/active system configuration
Be aware of the impact of SyncMirror:
Write performance will decrease in heavy load situation ~5%
Read performance can increase if you activate read from both plexes
Mirrored plexes result in half the usable max spindle count
Connections are important:
Stretch MetroCluster interconnect is IB with MPO Adapter – check customer’s patch panels
Remember to account for patch panels in distance and link budget calculations
Ensure that the right type of fiber is in use
NetApp Confidential - Limited Use

32. MetroCluster Complements Host Clustering
Solution Highlights
No SPOF
High failure resiliency—no 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 site-wide 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 configurations—fully 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.
NetApp Confidential - Limited Use

33. 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
NetApp Confidential - Limited Use

34. Multiple Subnet Support
\etc\mcrc
Configuring the network interfaces for the two nodes of a MetroCluster configuration historically has worked the same way as for a standard active-active configuration outlined in the Data ONTAP Active/Active and Network Configuration guides. Now, effective with Data ONTAP 7.3.2, each of the nodes of the MetroCluster configuration may reside on a different subnet. For those customers desiring this configuration, the following is required.
Set the option cf.takeover.use_mcrc_file to on.
Configure the /etc/mcrc file on each FAS controller.
The /etc/mcrc file, in conjunction with the cf.takeover.use.mcrc_file option, should be used on MetroCluster configurations in which the partner nodes reside on separate subnetworks. Normally, when a node (for example, nodeA) takes over its partner (nodeB), the system runs the partner's /etc/rc file and partner interfaces are configured locally to handle incoming traffic for the taken-over partner. The requires that the local and partner addresses are on the same subnetwork.
When the cf.takeover.use.mcrc_file option is enabled on nodeA, nodeA will use nodeB's /etc/mcrc file upon takeover, instead of nodeB's /etc/rc file. The ifconfig commands in the /etc/mcrc file can configure IP addresses on nodeA's subnetwork. With the correct ifconfig, virtual IP (VIP), and routing commands in the /etc/mcrc file, the resulting configuration allows hosts connecting to nodeB to connect to node A. for detailed instructions on creation of the /etc/mcrc file see the Active/Active Configuration Guide located on the NetApp NOW site.

35. NetApp Confidential - Limited Use
What’s New?
MetroCluster TieBreaker Solution
Vmware Certification with HA
Vmware Certification with FT
NetApp Confidential - Limited Use

36. 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)

37. NetApp Confidential – Limited Use
TieBreaker Diagram
NetApp Confidential – Limited Use

38. MetroCluster for VMware Environments
Delivers continuous data availability
Complements VMware® host environment with complete server to storage availability
Improves read performance
Interleave reads from mirrored arrays
Simplifies operations
Zero interdependencies
Deploys with confidence
Tested interoperability
Leverages “Virtualization Guarantee Program”
ESX Server
VMotion® Technology
MetroCluster
Many companies have employed server virtualization technologies to reap the benefits of increased server consolidation and utilization. And as more customers look to virtualization as a means to cost effectively manage their server environment, they want to extend virtualization to their business critical applications which also requires a higher level of data availability.
Our partnership with VMware provides a complete availability solution from server to storage. VMware HA provides application availability while MetroCluster provides continuous data availability.
Another key benefit in virtualized environments is the improved read performance. As a clustered solution, MetroCluster is able to read from both arrays for improved performance.
MetroCluster is a self-contained continuous availability solution with zero dependency on the OS platform and applications, simplifying operations. So, there’s no need for complex scripting or ongoing maintenance.
Because we work with VMware to test interoperability, and our track record of worldwide deployments in every major vertical market, you can deploy the solution with confidence.
Finally, you can leverage the virtualization guarantee program to sell the storage savings benefits.
NetApp Confidential - Limited Use

39. MetroCluster – Reference Stories
Lee County Court of Clerks – Public Sector, N.A
Full Story:
United Health Network – Healthcare, N.A
Full Story:
Bitburger Brewery Group – Retail, Germany
Full story:
New customer stories coming soon
NetApp Confidential - Limited Use 39

40. NetApp Confidential - Limited Use
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
Stretch MetroCluster
Data Center
NetApp Confidential - Limited Use

41. 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
NetApp Confidential – Limited Use

42. 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?
NetApp Confidential – Limited Use

43. SnapMirror Sync versus MetroCluster
Feature
SnapMirror Sync
MetroCluster
Replication Network
IP or FC
FC only
Limit on concurrent transfers
Yes
No limit
Distance limitation
Up to 200 KM (latency driven)
100 KM
Replication between HA pairs No
Manageability
FilerView, CLI, Ops Mgr
CLI
Multi-use replica
Yes (FlexClone)
No (dump)
Limited (app can read from both plexes)
Deduplication
Dedupe volume and sync volume cannot be in the same aggregate
No restrictions
Use of secondary node for async mirroring
No. Async replication has to occur from primary plex

44. 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
Controller sizing for MetroCluster is the same as a standard active/active system configuration
Be aware of the impact of SyncMirror:
Write performance will decrease in heavy load situation ~5%
Read performance can increase if you activate read from both plexes
Mirrored plexes result in half the usable max spindle count
Connections are important:
Stretch MetroCluster interconnect is IB with MPO Adapter – check customer’s patch panels
Remember to account for patch panels in distance and link budget calculations
Ensure that the right type of fiber is in use
NetApp Confidential – Limited Use

45. NetApp Confidential - Limited Use
References
TR MetroCluster Design and Implementation Guide
TR3517 – MetroCluster Upgrade Planning Guide
TR3552 – Optical Network Installation Guide
TR High Availability and Disaster Recovery for VMware using NetApp SnapMirror and MetroCluster
Data ONTAP Active/Active Configuration Guide
MetroCluster Compatibility Matrix
NetApp Confidential - Limited Use

46. NetApp Confidential - Limited Use
Value of MetroCluster
Data Center
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
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
To support the value prop, these are the key values of MetroCluster.
The key thing to keep in mind is that MetroCluster is the only enterprise vendor that offers a continuous availability solution that combines synchronous mirroring with array-based clustering. And we can do this at a lower cost with less complexity. The 3 value messages are simple, continuous availability and cost effective. Let’s look at the supporting value statements:
Simple: Most storage-based data mirroring solutions require 2 to 3 times the operational effort in terms of storage system administration, configuration, and maintenance because the replication relationships between the primary and secondary storage array and the host based clustering are all managed separately. This not only increases management overhead, it can also exposes you to greater risk if configuration inconsistencies arise between the primary and secondary storage arrays and the cluster configuration. Because NetApp MetroCluster is a true clustered storage solution, the active-active storage pair is managed as a single entity, eliminating duplicate administration work and maintaining configuration consistency.
Continuous Availability: NetApp MetroCluster enhances the built-in redundancy of NetApp hardware and software, providing an additional layer of protection for the entire storage and host environment. Whether your environment is composed of standalone servers, high-availability clusters, or virtualized servers, MetroCluster seamlessly maintains application data availability in the face of a total storage outage due to losses of power, cooling, or network connectivity; storage array shutdown; or operational error. This is because MetroCluster is an array-based active-active clustered solution, eliminating the need for complex failover scripts, server reboots, or application restarts. In fact, most MetroCluster customers report that their users experience no application interruption when a cluster recovery takes place.
From a cost perspective, MetroCluster is 50% lower cost. This includes lower software acquisition cost and cost of ownership of the solution due to its simple to manage architecture. But what makes the cost efficiency story even stronger is the added benefit of deduplication storage efficiency and the integration with server virtualization.
NetApp Confidential - Limited Use 46

47. 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.