According to a recent study by IDG’s International Data Center (IDC), increasing numbers of businesses today are moving to cloud and virtualization platforms.1 Why? Easier administration, faster disaster recovery and IT staff are freed up to address core business processes. After the initial cost, businesses believe they will save in the long run if they go virtual.
This virtual platform is most likely EMC’s VMware. But what happens if a VMware network goes down? Disaster recovery for this virtual platform requires specialized tools. VMware created several for this purpose, including its vCenter Site Recovery Manager.
Stages of Disaster Recovery with Site Recovery Manager
Backup of VMware will normally be in two parts: offsite backup (in case EVERYTHING goes) and onsite (available) replicates of machines and servers. This replication can be physical to virtual (P2V) or virtual to virtual (V2V).
If Recovery Site Manager is used, it plugs into the VMware console, and uses an Oracle or MS SQL server database to connect between the site to be protected and the recovery target. It automates the following processes.
DR Step One: Replication
Data can be replicated between the protection and recovery sites. VMware works with a ton of software developed just for this purpose: to create mirror images of the virtual machines. EMC, of course, wants you to use its own products:
- Symmetric: remote storage replication
- Celerra: asynchronous file system and iSCSI LUN replication for Celerra storage environments
- Mirrorview: synchronous or asynchronous local or remote mirroring of stored data
But do any online search for “replication software” and you can find decent software that will also do the job, compatible with VMware. The software helps Recovery Manager find arrays and machine images; it also helps automate failover and testing with Recovery Manager.
To keep things even simpler, EMC’s latest version of Recovery Manager works with vSphere 5.0 to create copies of VMs over the network. It uses the vSphere replication engine to copy only changed blocks to the recovery site (incremental copies).
A VM with one type of data storage can be sent to a target site using a different type.
Step Two: Create Protection Groups
This is the heart of Recovery Manager.
Recovery Manager allows an administrator to create groups containing individual virtual machines (VMs) for protection. Each VM (LUN) image file (mirror) is stored in a datastore. Recovery Manager then creates datastore groups from these datastores; the groups failover together in a DR situation.
- Assign individual datastores or groups to include in a protection group, and individually configure them for recovery activities.
- Assign each group a level of priority (from 1-5). The highest priority group starts recovery first and must finish before the next priority group of VMs begins operations
- Set VM dependencies within a priority group (i.e. required systems are started before VMs that depend upon them are started).
To work, Recovery Manager requires that VMware VirtualCenter and RM Server are installed at the protection and recovery sites, with sufficient storage space for recovery.
Step Three: Don’t forget about your physical footprint
EMC thought of this, too, and developed RecoverPoint. This is a network-based application that continuously protects the main server’s data. It works with all kinds of arrays, and can protect one server, or a whole server farm. RecoverPoint integrates completely with other VMware disaster recovery software.