This chapter is from the book
This chapter is from the book
This chapter is from the book
Disaster Recovery
Disaster recovery is an area of many organizations that has at least some, if not a lot, of room for improvement. When looking at a disaster recovery plan, the following things are important to consider:
- Data is available with an RTO that meets the business’s requirements to operate and the data is from a point in time that meets the RPO of the organization.
- Data has been verified as being valid.
- A runbook has been defined for how to and in what order to restore.
The first point is present in most organizations, whereas the second and third are not. It should not be surprising because a failure warranting declaring a disaster is not often needed. Nonetheless, a solid runbook should be defined for your infrastructure. A runbook for restoration for your virtual infrastructure is crucial; however, consider the back-end networking infrastructure first as your virtual machines will be of no use without it.
When looking at recovering your virtual infrastructure, the ideal setup is to replicate among your storage devices and use VMware’s Site Recovery Manager (SRM) to automate your restore. Site Recovery Manager is further discussed later in this chapter; however, for those not familiar, it assists in automating the recovery of virtual machine environments during a disaster.
You may also use the set of replicated data to manually configure the virtual machines and power them on at your disaster recovery location. Additionally, you can choose another method of manual restoration. This could be using a copy of the virtual machine files from some other mechanism or using a backup product to perform a bare-metal copy of the machine and restoring it to a newly configured virtual machine. For the purposes of this discussion, we talk in detail about the use of VMware’s Site Recovery Manager as it provides the best mechanism. Before doing that, though, the following sections talk briefly about the other options.
Manual Disaster Recovery
When looking at implementing a manual data recovery plan, you need to ensure you are doing a few things that Site Recovery Manager would be automatically handling or assisting with. Many times, the use of manual methods is the result of a lack of sponsorship of the initiative in terms of funding; however, that does not mean the process cannot work. If you are creating a manual data recovery plan consider the following.
- Ensure data is being replicated/copied and is current with your RPO.
- Ensure your processes for restoration meet your RTO.
- Ensure the Disaster Recovery (DR) site hardware is supported and will support the load in the event of a disaster.
- Ensure the recovery processes work by performing regular DR tests.
- Ensure the runbook is updated regularly as network, application, and other requirements change.
By keeping these points in mind, your disaster recovery efforts will be successful; however, you will have to perform many of the steps manually.
Whereas storage replication was previously a condition for using Site Recovery Manager, the latest version now supports host-based replication. If you were previously unable to use Site Recovery Manager because of the storage replication requirement, you should reevaluate the product with host-based replication.
Site Recovery Manager
VMware offers a product called Site Recovery Manager that helps automate most of the process of recovering virtual machines during a disaster. The product allows for isolated testing to ensure recovery is possible in the event of an actual disaster as well as the ability to failback in version 5.0.
When installed at both the production and disaster recovery locations, the product provides for a centralized approach to defining replication and recovery plans. In prior versions, SRM relied on the storage itself to perform the replications and integrated with the storage using a supported Storage Replication Adapter (SRA). This limited the product for some entities with supported storage. Even those with supported storage devices in both locations might not have had matching storage solutions and, hence, no supported replication infrastructure in place.
SRM 5.0, however, has expanded its market base with the introduction of vSphere Replication (VR). This allows replication from one location to another, regardless of the type of storage on both ends. One or both ends can even be local or directly attached storage. SRM is also protocol independent so you can replicate among Fibre Channel, iSCSI, or NFS storage.
For more information on Site Recovery Manager, check out Administering VMware Site Recovery Manager 5.0 by Mike Laverick. This book provides an in-depth discussion of the product, using it in a number of scenarios, and is a great read when defining a disaster recovery solution in a virtualized environment.