NetApp Business Continuity Solutions for VMware Virtualized Environments

Transcription

1 White Paper NetApp Business Continuity Solutions for VMware Virtualized Environments NetApp January 2011 WP-7125

2 TABLE OF CONTENTS 1 EXECUTIVE SUMMARY INTRODUCTION TECHNOLOGY OVERVIEW... 4 VMWARE VSPHERE PLATFORM... 4 VMWARE HIGH AVAILABILITY AND VMWARE FAULT TOLERANCE... 5 VMWARE DRS HOST AFFINITY... 7 VMWARE SITE RECOVERY MANAGER (SRM)... 7 NETAPP ENTERPRISE CLASS AVAILABILITY... 7 NETAPP METROCLUSTER... 8 NETAPP SNAPMIRROR... 9 NETAPP FLEXCLONE NETAPP SNAPMANAGER FOR VIRTUAL INFRASTRUCTURE (SMVI) USE CASE: MICROSOFT APPLICATION MIXED WORKLOAD IN A VMWARE ENVIRONMENT WITH NETAPP HIGH AVAILABILITY CONTINUOUS AVAILABILITY DISASTER RECOVERY CONCLUSION REFERENCE DOCUMENTATION NETAPP VMWARE LIST OF FIGURES Figure 1) VMware high availability Figure 2) VMware fault tolerance Figure 3) VMware and MetroCluster deliver continuous availability to 24x7 data center... 8 Figure 4) Comprehensive protection with disaster recovery...10 Figure 5) Microsoft application mixed workload in a VMware and NetApp HA configuration...12 Figure 6) Microsoft application mixed workloads in a VMware and NetApp MetroCluster configuration Figure 7) Disaster recovery with VMware and NetApp SnapMirror...14 Figure 8) SnapMirror relationship transparently moved to the surviving controller NetApp Business Continuity Solutions for VMware Virtualized Environment

3 1 EXECUTIVE SUMMARY Today s IT organizations are under relentless pressure to maintain continuously available infrastructures for their businesses while enabling the security, integrity, and confidentiality of valuable data. But they also need to be able to respond dynamically to changing needs, including unplanned events, such as power outages and natural disasters, and planned outages, such as software upgrades and the addition of new hardware. In today s economy, these requirements must all be met at the lowest cost possible. Whatever the cause, downtime and data loss can have a devastating affect on the bottom line. Revenue cannot be generated unless the business is up and running. Due to the increasing dependency of businesses on their IT organizations, businesses are scrutinizing various causes of outages and are adopting cost-effective systematic approaches to reduce the risks of downtime. As typical IT deployments range from single data centers to multiple data centers, protection might be needed in a single data center, across multiple data centers regionally located, over geographically dispersed data centers, or a combination of all three. NetApp and VMware address the above requirements by providing cost-effective solutions through virtualization. While today s x86 computer hardware was designed to run a single operating system and single application, virtualization lets you run multiple virtual machines (VMs) on a single physical machine. Each virtual machine shares the resources of that one physical computer across multiple environments. Built on a business-ready architecture, VMware is the market leader in server virtualization. VMware vsphere virtualizes and aggregates the hardware resources of your x86-based computer through its ESX and ESXi hypervisor architectures to create virtual machines capable of running their own operating systems and applications. Multiple operating systems safely run concurrently on a single physical computer, sharing hardware resources and maximizing your system s compute power. Consolidating multiple workloads on one physical computer helps optimize resource utilization, but the impact of downtime is much greater. To provide the same level of high-availability protection in the virtual server environment as in the physical environment, a VMware high-availability (HA) cluster can be configured by pooling together ESX/ESXi servers and the shared resources across a collection of hosts and virtual machines. Your HA cluster might be implemented with all hosts in a single data center or have added protection against single data center failures by stretching the cluster across multiple locations. Locations might be as close as separate floors of the same building, or locations stretched across a metropolitan area. In both configurations, this is a single cluster. Once a host is added to the cluster, the host s resources become part of the cluster s resources. Any cluster server may host a VM. Should a VM s host fail, the VM will fail over to another available host, and your mission-critical application will be automatically restarted. For protection beyond a localized region, VMware s Site Recovery Manager (SRM) automates the recovery of virtual machines between sites that might be thousands of miles apart. In a typical SRM configuration, the primary or protected site runs your mission-critical applications and services. The recovery site provides an alternate environment should your protected site fail. Making sure of data availability is of paramount importance to NetApp. NetApp storage is designed with high availability, flexibility, and efficiency in mind. A suite of capabilities within the NetApp FAS platform protects against component failures and even entire system or data center failures to keep your critical business operations running. These functions work in tandem with NetApp storage efficiency technologies to reduce capacity and operational costs so that you can provide protection for more of your environment. To increase coverage, NetApp MetroCluster enables clustered storage to be stretched over two locations, providing protection from typical data center failures such as local power outages, loss of the network, hardware and software failures, and even human error. The two locations can be as close as two different computer rooms in the same building or as distant as two separate sites up to 100km apart. Since this is a single cluster, your mission-critical applications continue to run even if one of the locations fails. 3 NetApp Business Continuity Solutions for VMware Virtualized Environment

4 However, MetroCluster cannot provide protection against a complete regional disaster. For more comprehensive protection against more widespread disasters, a DR site should be placed at a much greater distance from the primary sites. NetApp SnapMirror software uses mirroring methodology that spans unlimited geographic distances. SnapMirror builds on the NetApp Snapshot technology to perform thin replication efficiently identifying, compressing, and replicating only changed blocks. Implementing a combination of MetroCluster for continuous availability and SnapMirror for disaster recovery provides complete business continuity. In a nutshell, VMware and NetApp technologies directly complement each other for a highly available cost-effective virtualized IT infrastructure. On the server side, VMware s virtualization technology uses commodity servers to create flexible, scalable architectures that maximize server resource utilization. VMware s HA clustering feature provides high availability for your server environment, and its SRM disaster recovery capabilities provide protection from regional outages. On the storage side, NetApp works in tandem with VMware to extend the architecture by adding shared storage with such features as MetroCluster for continuous data availability and SnapMirror for data protection from regional disasters. 2 INTRODUCTION Competitive market trends drive enterprises to provide high levels of business continuity, while massive data growth, economic conditions, and physical data center limitations place extreme pressure on traditional IT infrastructures. As a result, businesses are turning to virtualization to manage cost and reduce complexity, while employing high availability and fault tolerant technologies for business continuity. Two industry leaders VMware and NetApp combine server and storage virtualization to deliver simple, cost-effective, robust continuous-availability and disaster recovery solutions. VMware high availability and fault tolerance provide your virtualized server environment with uniform, cost-effective failover protection against hardware and software failures. NetApp storage and its MetroCluster and SnapMirror solutions further extend the architecture of your virtualized environment with management simplicity, growth, continuous availability, and data protection by seamlessly providing shared network storage to the server environment. The combination of these technologies offers a great value proposition for the entire infrastructure of your environment, enabling you to effectively manage both planned and unplanned downtimes of your mission-critical applications. Through a typical deployment scenario with mixed Microsoft application workloads, this white paper provides an overview of a virtualized infrastructure that deploys VMware with NetApp. 3 TECHNOLOGY OVERVIEW This white paper is not intended as a comprehensive document on the VMware vsphere platform. However, a high-level discussion of several of its features is included to provide context for the configurations and examples included here. VMWARE VSPHERE PLATFORM vsphere is VMware's enterprise-class virtualization platform for a virtual infrastructure. Key components of the vsphere platform include: Infrastructure services to virtualize and aggregate server, storage, and network resources, then allocate them to applications on demand. Infrastructure services include: - ESX and ESXi hypervisor architectures, which abstract server hardware resources, allowing multiple virtual machines (VMs) to share those resources - Distributed Resource Scheduler (DRS), which aggregates resources across multiple physical systems into a cluster and allocates those resources on demand to virtual machines in the cluster - DRS Host Affinity, which allows you to set policies to govern VM movement across servers within the HA cluster 4 NetApp Business Continuity Solutions for VMware Virtualized Environment

5 Application services that provide service levels such as availability, security, and scalability, including: - VMware VMotion, which enables the live migration of VMs across physical servers - VMware high availability (HA), which provides cost-effective, automated restart of an application if the VM on which it is running fails as a result of hardware or operating system failure - VMware fault tolerance (FT), which delivers continuous availability to an application running on a guest VM without any data loss or downtime due to hardware failure The VMware vcenter Server provides unified management of all hosts and VMs in a data center so that you can easily manage your virtualized environment from a central location. Site Recovery Manager (SRM) is a plug-in to vcenter that leverages array replication between a primary (protected) site and a DR (recovery) site to automate tasks for disaster recovery with the push of a button VMWARE HIGH AVAILABILITY AND VMWARE FAULT TOLERANCE A key philosophy within VMware is to provide a full range of high-availability features in its virtualization platform. Two such features which provide additional protection to complement built-in hardware protection are VMware high availability (HA) and VMware fault tolerance (FT). With VMware HA, a set of ESX or ESXi host servers is combined into a cluster with a shared pool of resources to provide an easy-to-use cost-effective high-availability infrastructure for virtual environments. HA protects against both host and VM guest failures by continuously monitoring all virtualized servers in a resource pool and detecting physical server and operating system failures. If a host server fails, HA makes sure requisite resources are available and automatically restarts the affected virtual machines on other servers in the pool. HA also protects the VMs in the cluster against operating system failures by monitoring heartbeat information of each VM. If a VM fails, it is automatically restarted by HA on the same host server. VMware HA provides affordable uniform high availability across the virtualized environment without the cost and complexity of a failover solution that is tied to either an operating system or a specific application. Figure 1) VMware high availability. VMware fault tolerance uses VMware vlockstep technology to provide operational continuity with simplicity and low cost. Requiring neither custom hardware nor software, VMware FT takes high availability to the next level by building it directly into the x86 hypervisor to deliver hardware style fault tolerance to virtual machines. Whereas VMware HA automatically restarts a failed VM, VMware FT eliminates the downtime associated with a hardware failure. FT enables an affected VM to transparently fail over to a different host server with no disruption of service when its host fails. FT works by creating a 5 NetApp Business Continuity Solutions for VMware Virtualized Environment

6 live shadow instance of the primary on a different server; then vlockstep maintains synchronization between the servers by recording events on the primary instance and transmitting them to the secondary for replay. At any given time, both VMs play the same set of events. Although both the primary and secondary VMs execute the same set of instructions in the same sequence, including I/O operations, only the outputs of the primary VM take effect. The two VMs constantly heartbeat against each other, and if either instance loses the heartbeat, the other takes over immediately. Automatic detection of failures and seamless failover make sure that applications continue to run without interruptions, user disconnects, or data loss from hardware failures. Figure 2) VMware fault tolerance. The high-availability feature of an HA cluster requires shared storage. NetApp s unified storage architecture supports all storage protocols that can be deployed in VMware environment, including Fibre Channel, iscsi, and NAS protocols. Among various storage protocols that can be used on the vsphere platform, NAS/NFS is the most costeffective and simplest option. With NFS, storage management tasks are all done within NetApp controllers, significantly simplifying such arduous tasks as LUN management. The manageability of the whole environment is simplified with clear roles and responsibilities, freeing IT personnel to focus on other tasks. The files corresponding to the virtual disks of the VMs stored in the shared storage pool are thin provisioned by default and also deduplicated. Deduplication technology from NetApp assists VMware deployments to eliminate duplicate data in their environments, enabling greater storage utilization on production environments. NetApp deduplication technology enables multiple virtual machines to share the same physical blocks of storage. Deduplication can be seamlessly introduced into a virtual infrastructure without having to make any changes to VMware administration, practices, or tasks. For the largest storage savings, NetApp recommends grouping similar operating systems and similar applications into the same deduplication-enabled volume. This leads to very high utilization of storage as well as drastic reduction in the total amount of storage required. 6 NetApp Business Continuity Solutions for VMware Virtualized Environment

7 VMWARE DRS HOST AFFINITY DRS (Distributed Resource Scheduler) Host Affinity was introduced with VMware vsphere 4.1 to facilitate management of VM movement across servers in the HA cluster. This feature is especially useful with VMware HA and FT, as it allows you to tune your environment by setting granular policies to control the movement of your VMs. Prior to the advent of host affinity, the host to which a VM was relocated could not be easily determined or controlled. This made it possible, in a stretch cluster configuration, for a VM to be moved to a host that was not local to the same data center as the storage containing that VM s data. But with DRS Host Affinity, you have the ability to limit a VM to a subset of servers within the pool. For example, you can define different groups of hosts in the cluster and assign a subset of your VMs to one of these groups. This feature offers significant benefits, as you might need to base movement of your VMs on such attributes as locality, availability, performance, licensing requirements, and so on. VMWARE SITE RECOVERY MANAGER (SRM) While VMware HA provides high-availability protection and VMware FT provides continuous availability, this coverage is regional. Your environment might still need protection against a regional disaster, but deploying and maintaining a disaster recovery solution over a longer distance can be very challenging. For instance, one of the most time-consuming parts of DR failover in a VMware environment is the execution of steps necessary to connect, register, reconfigure, and power up virtual machines at the DR site. VMware addresses this with its vcenter Site Recovery Manager (SRM). SRM, which is a plug-in to vcenter, enables two separate vcenter environments to communicate with each other by integrating with their network and storage layers. While SRM does not actually replicate the storage itself, VMware partners with storage providers such as NetApp to deliver this functionality. The storage provider provides an SRM adaptor to be installed with SRM to enable communication between the SRM server and the vendor s shared storage. Through the storage provider s data replication facility, datastores containing VMs are replicated from the protected (primary) site to the DR (recovery) site. VMs to be failed over are grouped together under SRM protection. Your IT specialist provides the set of steps needed to recover the protected environment, including any special instructions such as application startup order, to the SRM server. You now have protected VMs that are replicated in datastores across storage arrays, knowledge of the datastores, and startup instructions for the VMs and their applications. So with the push of a button, your disaster recovery plan that is powered by SRM can automate such tasks as: Quiescing and breaking the relationship between the storage arrays at the protected and recovery sites Connecting the replicated datastores to the ESX hosts at the DR site Powering off specified VMs at the DR site to free up resources for the VMs migrating from the protected site Reconfiguring VMs as defined for the network at the DR site Powering on VMs in the order as defined in the recovery plan Executing any custom commands that have been stored in the recovery plan Configuring VMs into SRM protection groups with shared resources not only enables them to be recovered together, but also enables setup automation and testing of disaster recovery plans. With SRM, downtime resulting from a disaster is significantly reduced. NETAPP ENTERPRISE-CLASS AVAILABILITY Starting with NetApp s unified storage platform, capabilities are built directly into the storage subsystem to provide resiliency, protection against data loss, and proactive monitoring and maintenance to deliver enterprise-class availability. Component redundancies include the power supply, storage HBA, and Ethernet NIC. NetApp RAID-DP provides protection against failure of up to two disks in a single RAID 7 NetApp Business Continuity Solutions for VMware Virtualized Environment

8 group. Multipath HA supports multiple paths to all storage shelves, providing controller access to all shelves in a stack. A NetApp HA pair controller configuration pairs two storage controllers together as partners so that each controller is cabled to all of the drive shelves in the storage system. The controllers heartbeat against each other, and if one fails, data service is transferred to the surviving partner. When a dual-controller system is deployed, nondisruptive upgrades can be performed. NETAPP METROCLUSTER MetroCluster builds on NetApp s built-in high-availability capabilities to add its unique array-based clustering solution with synchronous data mirroring for greater protection and availability of mission-critical data. The clustering feature enables continued access in the event of any single component failure such as loss of disk shelf, disk loop, controller, and so on. If the controller fails in one location, the partner controller at the second location will detect the failure and automatically assume the responsibility of the failed controller. By combining MetroCluster with VMware technologies, you can deploy a comprehensive solution to address both application and data availability. Figure 3) VMware and MetroCluster deliver continuous availability to 24x7 data center. MetroCluster provides automatic and transparent recovery for any failure within the data center. It also enables transparent hardware and software upgrades to the storage array by redirecting data access to the remote site during the upgrade process. Thus, MetroCluster enables zero planned and unplanned downtime with no data loss and an RPO/ RTO of zero so there is no disruption to applications, clients, or users. Unlike traditional remote mirroring solutions, which require complex storage failover scripting, MetroCluster enables storage recovery from site failures within minutes with a simple single-command failover. Once the storage fails over to the second site, the identity of the source storage controller also fails over with it. The VMware environment continues to connect to the same storage target unaware that the shared storage is now hosted on a different array. 8 NetApp Business Continuity Solutions for VMware Virtualized Environment

9 MetroCluster replicates synchronously between sites. MetroCluster data replication doesn t involve both controllers; the primary controller simultaneously writes to both mirror plexes in the same aggregate. In an environment that has a high ratio of random reads, MetroCluster improves read performance by up to 80% with read operations occurring on both storage plexes. MetroCluster works with NetApp s unique deduplication technology where the mirror write I/O is automatically deduped on both the primary and secondary arrays for added savings. Whether you have a single data center or multiple data centers in a campus or metropolitan area, MetroCluster is a costeffective solution that can provide the continuous availability and data protection you need to keep your environment going. Configuring MetroCluster data replication is extremely simple. There is no configuration of source and target devices, or pairing relationship. The mirroring is done transparently to the host. The host only sees primary volumes, and this further simplifies the management of the volumes at the host level. MetroCluster is an easy-to-administer solution. You simply set it once, and MetroCluster does all the work for you. It also enables you to eliminate repetitive change management activities. Any changes you make to the primary storage are automatically reflected to the secondary storage. This not only saves time and simplifies management, it also helps reduce exposure to human error, a major cause of downtime according to leading industry analysts. While deploying VMware HA and FT with MetroCluster offers several advantages, concerns might arise around where a VM runs relative to its storage. As such, MetroCluster is an excellent use case for HA clustering with DRS Host Affinity. Consider the scenario where an HA cluster spans MetroCluster sites so that the ESX host servers are geographically dispersed over two data centers. Upon host failure, affected VMs will be assigned to other available hosts, which might not be in the same data center as their storage. However, to avoid unwanted latency, you might want to restrict VM movement to the same data center as its storage. DRS Host Affinity allows you to specify a subset of nodes within the cluster which might run a VM, as opposed to allowing the VM to migrate to any available node in the cluster. This feature is especially useful in a MetroCluster environment, because the ESX servers can span data centers that are up to 100km apart. And Host Affinity is not limited to the HA feature in DRS. All VMs in the cluster, both FT and non-ft, can benefit from DRS Host Affinity. FT-enabled virtual machines can be configured to take advantage of DRS functionality for load balancing and initial placement. NETAPP SNAPMIRROR SnapMirror software is the value leader in the industry when it comes to disaster recovery (DR). Its simplicity and flexibility make it an affordable DR solution. SnapMirror replicates data over LAN or WAN to a remote site, providing you with high data availability and fast disaster recovery for your mission-critical applications in both traditional and virtual environments. SnapMirror supports synchronous replication, making sure of zero data loss; semisynchronous replication that supports recovery point objectives (RPOs) in seconds with minimal impact on the host application; and asynchronous replication that can meet RPOs ranging from 1 minute to 1 day. SnapMirror provides thin replication and works with NetApp deduplication technology. It replicates data in deduplicated state so that network usage is minimal, and space savings from deduplication on primary storage are automatically delivered on the secondary storage system. The SnapMirror network compression feature enables data compression, further reducing network bandwidth and cost. SnapMirror complements MetroCluster to further extend the protection of data from regional disasters. The multisite solution is desirable if your business requires zero data loss. Using MetroCluster to synchronously mirror data between two locations within a building, campus, or metropolitan area and SnapMirror to asynchronously replicate data across long distances, you get the best of both worlds: seamless failover capabilities with zero data loss over metropolitan areas and disaster protection over wide areas. 9 NetApp Business Continuity Solutions for VMware Virtualized Environment

10 With your data replicated to your DR site, you need to resume operations as soon as possible following a disaster. VMware vcenter SRM migrates the protected inventory of protected VMs to the DR site, accesses the data replica at the DR site, and brings up the failed VMs to reinstate your missioncritical applications. Your DR environment is now ready to provide the same services as the primary site. By using SnapMirror to mirror data to one or more NetApp storage systems, you can use the mirrored data wherever and whenever you need it. SRM not only facilitates disaster recovery, it also simplifies such tasks as DR testing. To maximize its usage, your DR site can easily serve multiple purposes, including: Disaster recovery Off-site tape archival (replicate to remote site and then backup to tape) Online data migration Testing, development, QA, and production staging (combined with FlexClone ) Figure 4) Comprehensive protection with disaster recovery. NETAPP FLEXCLONE You can extend the value of your disaster recovery site with NetApp FlexClone technology, which enables cloning of your DR data instantly generating replicas of data volumes and datasets without requiring additional storage space at the time of creation. Each cloned volume is a transparent, virtual copy that you can use for essential enterprise operations, such as testing, bug fixing, platform and upgrade checks, multiple simulations against large datasets, remote office testing and staging, and market-specific product variations, all without affecting your production operations. When NetApp FlexClone technology is combined with SnapMirror and SRM, testing the DR solution can be performed quickly and easily, requiring very little additional storage, and without interrupting the replication process. When this functionality is used, an additional copy of the data is not required. For example with a 10GB LUN, another 10GB LUN isn t required. Only the metadata is required to define the LUN. FlexClone volumes only store data that is written or changed after the clone was created. FlexClone allows a read/write copy of the DR data to be created in a matter of seconds, requiring only a small percentage of additional capacity for writes that occur during testing. FlexClone volumes share common data blocks with their parent FlexVol volumes but behave as independent volumes. This allows DR testing to be completed without affecting the existing replication processes. Testing of the DR environment can be performed, even for an extended period of time, while replication is still occurring to the parent FlexVol volume in the background to maintain the existing RPO. 10 NetApp Business Continuity Solutions for VMware Virtualized Environment

11 NETAPP SNAPMANAGER FOR VIRTUAL INFRASTRUCTURE (SMVI) SnapManager for Virtual Infrastructure (SMVI) leverages NetApp s Snapshot technology to create pointin-time copies of your data. SMVI accommodates backups of large datastores or virtual machines and allows restore at any desired granularity, including entire datastores, individual virtual machines, virtual machine disks (vmdks), or individual files within a guest OS. SMVI provides local backup and recovery capability with the option to replicate backups to a remote storage system using SnapMirror. This is all done on the storage system with no impact to the server environment. 4 USE CASE: MICROSOFT APPLICATION MIXED WORKLOAD IN A VMWARE ENVIRONMENT WITH NETAPP The following use case virtualizes a mix of Microsoft application workloads, including SQL Server, Exchange Server, and SharePoint Server on a VMware vsphere 4.1 virtual infrastructure as follows: Microsoft Exchange: 6 VMs (one mailbox server + its shadow instance, two hub servers, and two client access servers [CASs]) Microsoft SharePoint: 4 VMs (two web servers, one index server, and one DB server) Microsoft SQL Server: one VM (SQL Server) In addition to the Microsoft applications, the vcenter server is running under HA protection. The VMware vcenter Server Heartbeat feature is used on a pair of VMs with an active instance on one node and its standby (passive) instance on another node. HA host-based clustering and NetApp storage clustering technologies enable continuous availability features, and disaster recovery architecture provides protection against losses resulting from failures ranging from single VM crashes to sitewide outages. For simplicity, our example has the following business requirements: services are most critical to the success of the enterprise. To provide the required coverage, Microsoft Exchange is protected by VMware fault tolerance. Applications with less stringent availability requirements, including SharePoint and SQL Server, are protected by VMware high availability. In this example, our applications do not require application-specific backup and restore agents, so NFS-based shared storage was chosen for its simplicity and flexibility. SMVI accommodates backup and restore and works with SnapMirror for remote data replication. HIGH AVAILABILITY Figure 5 shows an HA cluster configured with three VMware ESX 4.1 hosts (nodes 1-3) with VMs on NetApp shared storage running the mix of Microsoft applications described in the previous section. The workloads are distributed as follows: Node 1 (4 VMs): Exchange Server applications (mailbox server, hub server, and CAS server) and a SharePoint Web server. Node 2 (4 VMs): Exchange Server applications (the mailbox shadow instance, hub server, and CAS server) and a SQL Server. Since the shadow instance of the mailbox server is running on this node for FT protection, node 2 will automatically take over if node 1 fails, and another shadow instance will be instantiated on node 3. Node 3 (3 VMS): SharePoint Server applications (DB server, index server, and Web server 2). A VM is running vcenter on node 2 with its standby on node NetApp Business Continuity Solutions for VMware Virtualized Environment

12 Figure 5) Microsoft application mixed workload in a VMware and NetApp HA configuration. At the server level, HA clustering protects the virtualized server environment from operating system and hardware failures with VMware HA and VMware FT. HA provides protection against operating system failures as well as hardware failures. If either SQL Server or SharePoint services go down because their respective VM fails, VMware HA will attempt to restart it on the same server. However, if the ESX server suffers a hardware failure, the VM will fail over, and services will be restarted on a different host server. For example, if node 2 goes down, the SQL Server VM will restart on one of the other two nodes. In our example, Microsoft Exchange requires greater protection so that its VM is configured with VMware fault tolerance for maximum coverage against host hardware failures. FT has created a live shadow instance on node 2 of the mailbox server VM hosted on node 1. Once created, the instance on node 2 remains active, accessing common shared storage, and receiving and processing the same events as the primary instance. If the primary instance fails, the instance on node 2 will immediately take over with no disruption in service to the users. Another shadow instance will be created on node 3 to protect the instance on node 2. At the storage level, the applications datastores are protected with NetApp paired HA controllers and built-in highly resilient features such as: - Built-in redundancy protecting from failure of power supply, fan, or disk controller - Multipath HA protecting from single disk path or port failure - RAID-DP protecting from single or dual-disk failure - Snapshot copies protecting from accidental erasure or destruction of data Although the environment is made highly available with clustered servers and a NetApp storage array with robust HA attributes, the data center itself is still a single point of failure (SPOF). A fire, flood, or power outage can take the whole data center down. You can provide protection against a sitewide outage 12 NetApp Business Continuity Solutions for VMware Virtualized Environment

13 with the deployment of a second location, which might be in another computer room of the same building as the original data center or might be a totally separate site up to 100km away from the first site. CONTINUOUS AVAILABILITY MetroCluster extends the coverage of high availability to continuous availability by adding a second location with synchronous data replication. For this configuration, the cluster has been spread across two buildings, and a fourth node has been added so that we now have two nodes per building with the following workload distribution: Building 1 - Node 1 (4 VMs): Exchange Server applications (mailbox server, hub server, and CAS server) and a SharePoint Web server - Node 2 (3 VMs): Exchange Server applications (the mailbox shadow instance providing FT protection, hub server, and CAS server) Building 2 - Node 3 (2 VMs): SharePoint Web server and SQL Server - Node 4 (2 VMs): SharePoint Server applications (DB server and index server) The mailbox server VM is configured with FT so that if node 1 fails, its application will run in the same data center on node 2. If both nodes 1 and 2 fail, the mailbox server VM will fail over to the second building. Since the HA cluster is spread across two locations, DRS Host Affinity rules are configured for the VMs running SQL Server and SharePoint to govern the VM movements such that if node 3 fails, VMs running the SQL Server and SharePoint Web server on node 3 will migrate to node 4 before attempting to move to either node 1 or 2 in the other building. vcenter is running on node 2 in building 1 with its standby on node 3 in building 2. Figure 6) Microsoft application mixed workloads in a VMware and NetApp MetroCluster configuration. 13 NetApp Business Continuity Solutions for VMware Virtualized Environment

14 While VMware cluster with HA and FT protects the server side, NetApp MetroCluster provides continuous data availability through its unique combination of array-based clustering with synchronous mirroring for zero data loss. MetroCluster provides automatic and transparent recovery for any failure within the data center. And configuring MetroCluster data replication is extremely simple. There is no configuration of source and target devices, or pairing relationship. The mirroring is done transparently to the host. The host only sees primary volumes, and this further simplifies the management of the volumes at the host level. NetApp MetroCluster works seamlessly with vsphere, providing continuous access to the shared storage containing all datastores hosting the VMs. DISASTER RECOVERY Although the environment is made highly available with VMware clustering and NetApp s storage array with robust HA attributes, you might still need protection against a regional disaster. In this situation, the most comprehensive disaster recovery plan is to have a third site that is further away from the other two sites. Adding to our use case, our third data center looks very similar to the single data center, with 3 hosts (nodes 5, 6, and 7) configured in an HA cluster on shared storage as depicted in the following diagram. Under normal conditions, this cluster hosts a variety of VMs running such applications as training and reporting using FlexClone data replicas. However, along with its normal operations, the third site is also capable of serving as a DR site with the aid of: NetApp SnapManager for Virtual Infrastructure (SMVI) to create point-in-time local copies of the datastores NetApp SnapMirror to replicate the datastores from the (primary) protected site to the (DR) recovery site VMware vcenter Site Recovery Manager to perform the failover of the SnapMirror relationship between the protected and recovery sites, then initiate activities to free up compute resources at the recovery site and bring up protected VMs NetApp FlexClone to create storage replicas to be used for such tasks as DR testing, data mining, and test and development Figure 7) Disaster recovery with VMware and NetApp SnapMirror. SMVI creates point-in-time on-disk backups of the datastores and initiates SnapMirror updates for replication of the datastores to the DR site. NetApp SnapMirror provides long-distance data replication from the primary site to the DR site. Operating either asynchronously or synchronously, SnapMirror utilizes NetApp Snapshot copies to make replication both easy and efficient. 14 NetApp Business Continuity Solutions for VMware Virtualized Environment

15 In our three-site configuration, while MetroCluster is used to protect data from site failure within a campus or metropolitan area, SnapMirror asynchronous is used for wide-area disaster protection across virtually any distance. As shown in Figure 8, the diagram on the left shows a three-site configuration in normal running condition. The diagram on the right indicates that when the primary site (site 1) fails, the operations are recovered at site 2 as a part of MetroCluster. After the storage takeover command is executed at site 2 and the storage controller at site 2 takes over the operation, the SnapMirror relationship to the DR site controller is transparently moved over to site 2 controller. This illustrates the simplicity of managing NetApp data replication in a multisite configuration. Figure 8) SnapMirror relationship transparently moved to the surviving controller. If both sites 1 and 2 fail, the DR site will need to assume operations. Microsoft server applications will need to run at the DR site. Since the DR site is a distinctly separate cluster, VMware Site Recovery Manager (SRM) facilitates communication between the two separate environments. All three Microsoft servers have been grouped under SRM protection to prepare for such a situation. In our example, the most critical application is Exchange, so it needs to be recovered first and foremost. SharePoint has a dependency on SQL Server such that SQL Server must be recovered before SharePoint. So the disaster recovery plan will specify the order of startup as Exchange, SQL Server, then SharePoint. SRM will: Through the NetApp SRM adapter, break the SnapMirror relationship to the storage at the primary site Present the datastores to nodes 5-7 Using site recovery plan, halt identified VMs that are running at the DR site to free up compute resources, then bring up the failed VMs and restore Microsoft Exchange, SQL Server, and SharePoint applications and services respectively to bring your business back to an operational state Although the primary goal of the DR site is to provide recoverability of your mission-critical environment in the face of disaster, it doesn t have to be an idle standby site. To get the most out of your environment, additional VM guests can be created for different applications. For example, the same set of hardware used for DR can also be use for training or reporting applications. It can be used to house a development and test environment. NetApp FlexClone technology can be used to create multiple, space-efficient copies of the data without disrupting the active environment or consuming a lot of storage space for these purposes. If there are enough CPUs and memory, all VM guests can even run simultaneously. It is really up to you, as your disaster recovery plan should be based on your specific situation to dictate which VMs stay active in the face of a disaster. 15 NetApp Business Continuity Solutions for VMware Virtualized Environment

16 5 CONCLUSION VMware + NetApp offer robust, cost-effective business continuity solutions for your virtualized IT environment by seamlessly integrating VM management with storage management. The unique offerings provide coverage across a broad spectrum of RPO and RTO requirements to meet your business needs, while significantly lowering your TCO when compared to hardware-based business continuity solutions. Virtualizing your environment increases its flexibility so that it can be quickly adapted to changing business needs, and reducing complexity with integrated management tools provides efficient operational simplicity. The VMware vsphere platform is the industry s most complete and robust virtualization platform, virtualizing business-critical applications with dynamic resource pools for unprecedented flexibility and reliability. NetApp provides a unified storage architecture suite of products and solutions, including MetroCluster and SnapMirror, to extend the flexibility and reliability of your IT solution and to provide premier data protection. NetApp MetroCluster is the only enterprise storage technology that combines synchronous mirroring with array-based clustering. It enables more robust business continuity to VMware environments by seamlessly providing continuous data availability across its shared storage infrastructure. MetroCluster offers many benefits, including: Protection of data from total storage failure due to sitewide outage. Improved storage efficiencies by leveraging NetApp deduplication technology. Read performance improvements up to 80%. The ability to apply hardware and software upgrades to the storage array without disrupting the environment. For example, if the primary site is unavailable, MetroCluster automatically redirects data access to the remote site. Since the storage at the remote site is a complete replica of the primary site, processing is uninterrupted. While MetroCluster provides continuous availability from local and/or sitewide failures, NetApp SnapMirror offers disaster recovery protection against large-scale disasters that affect your environment. SnapMirror offers the following benefits: Seamless to application Supports database and nondatabase data Provides native network compression to reduce network costs Enables business intelligence, test and development operations Easy testing of DR functions 6 REFERENCE DOCUMENTATION NETAPP NetApp High-Availability Solutions MetroCluster Design and Implementation Guide SnapMirror Async Overview and Best Practices Guide VMware vcenter Site Recovery Manager in a NetApp Environment A Continuous-Availability Solution for VMware vsphere and NetApp 16 NetApp Business Continuity Solutions for VMware Virtualized Environment

17 NetApp Solutions Guide: Microsoft Exchange Server, SQL Server, and SharePoint Server Mixed Workload on VMware vsphere 4, NetApp Unified Storage, and Cisco Nexus Unified Storage Disaster Recovery of Microsoft Exchange, SQL Server, and SharePoint Server Using VMware vcenter Site Recovery Manager, NetApp SnapManager and SnapMirror, and Cisco Nexus Unified Fabric NetApp and VMware vsphere Storage Best Practices VMWARE VMware High Availability Automating High-Availability (HA) Services with VMware HA VMware Fault Tolerance Protecting Mission-Critical Workloads with VMware Fault Tolerance VMware Fault Tolerance Recommendations and Considerations on VMware vsphere 4 VMware vsphere 4 Fault Tolerance: Architecture and Performance VMware Fault Tolerance Real-World Use Cases NetApp provides no representations or warranties regarding the accuracy, reliability or serviceability of any information or recommendations provided in this publication, or with respect to any results that may be obtained by the use of the information or observance of any recommendations provided herein. The information in this document is distributed AS IS, and the use of this information or the implementation of any recommendations or techniques herein is a customer s responsibility and depends on the customer s ability to evaluate and integrate them into the customer s operational environment. This document and the information contained herein may be used solely in connection with the NetApp products discussed in this document. Copyright 2011 NetApp, Inc. All rights reserved. No portions of this document may be reproduced without prior written consent of NetApp, Inc. Specifications are subject to change without notice. NetApp, the NetApp logo, Go further, faster, xxx, and xxx are trademarks or registered trademarks of NetApp, Inc. in the United States and/or other countries. <<Insert third-party trademark notices here.>> All other brands or products are trademarks or registered trademarks of their respective holders and should be 17 NetApp Business treated Continuity as such. Solutions WP for VMware Virtualized Environment