DR Checklist: How to Implement Effective Disaster Recovery in Virtualized Data Centers
Disaster Recovery Situational Analysis
Enterprises of all sizes are deploying applications on virtualized IT infrastructures and clouds, and reaping the benefits of ease of management, simplicity, flexibility, mobility and cost-effectiveness. This means that it should be equally commonplace for business continuity and disaster recovery efforts to be virtualized. However, this is not necessarily what is happening. Traditional replication and DR solutions were not developed to work within virtualized data centers, adding cost and complexity to the virtual infrastructure, with additional management tools and expertise needed to orchestrate disaster recovery. These organizations are tackling DR in a number of different, somewhat inefficient ways.
For example, organizations are utilizing:
- Local Data Protection / Backup – This involves the use of multiple products including a backup server, backup software, backup agents, snapshot software and a backup media. These devices are required to deliver a copy of the application data locally to an organization’s IT department. Typically these processes are run at night to provide IT with the opportunity to recover to yesterday’s data set, which is not acceptable for a mission critical application. While many of these backup tools work within virtual environments, they are focused on secondary copies of data, not recovery.
- Snapshot-based Replication – In these scenarios, replication is performed by taking periodic snapshots and subsequently transferring changed blocks of data over the network to a remote site. With some of these these tools, virtual-awareness is high, however replication features are sub-par. Snapshotting is a drain on production environments, with higher recovery time objectives (RTOs) and recovery point objectives (RPOs) and no application consistency grouping or point-in time recovery, rendering these solutions irrelevant for business-critical applications. Snapshot-based solutions are also weak on networking and automation features.
- Array-based Replication – These products are provided by storage vendors and deployed as modules inside the storage array. Array-based replication is oblivious to virtual IT configuration changes and does not support advanced VMware features, such as Storage DRS and Storage VMotion. This creates inefficiencies and even negates the benefits of virtualization, making DR manual and more complex, and rendering DR to the cloud nearly impossible. These solutions are not automated and require numerous points of control, multiple teams and duplicate management consoles, creating complexity and room for error.
- Host/Guest-based Replication – In these solutions a replication engine is installed on each individual virtual server. Such solutions do support heterogeneous storage which makes them less expensive to implement than storage-based replication, however installing an agent on each host/guest increases operational costs, server overhead and greatly limits scalability in large environments. Additionally, it takes away processing power from the application that is running your business.
The overarching problem with replication that takes place outside the virtualized infrastructure is that it undermines the inherent benefits of virtualization. Mission-critical applications can be effectively virtualized and managed in this scenario, but they cannot be effectively protected in a virtualized environment. This means that organizations are either keeping mission-critical applications on physical infrastructure or are moving mission-critical applications to a virtualized environment but not protecting them effectively.