Worst-Case Scenario Data Recovery: Replication Comes to the Fore
With emerging trends in data replication, credit unions now have a cost-effective, efficient and more technically advanced alternative to current tape-based disaster-recovery scenarios.
"With replication, companies can synchronize data between production and recovery centers, thereby eliminating or significantly limiting data loss in the event of system or entire site failure," Forrester Research analysts Rachel Dines and Stephanie Balaouras said in a new report.
Switching from, or at least supplementing, traditional tape backups has been taking place among credit unions in increasing numbers, and the firm's report, "The Past, Present and Future of Replication," said it found that 61% of American enterprises and 64% of European enterprises list upgrades to their disaster-recovery capabilities among their critical or high priorities.
The authors said the first step in moving to an advanced recovery solution is analysis of the IT environment: What hosts need protection? How does storage infrastructure map to the available replication methods? How much of the data needs replication? What availability requirements must a recovery systems meet?
Another important consideration is the optimal number of data-recovery strategies and products. Dines and Balaouras said too few can leave an IT environment vulnerable, while too many make it difficult to recover mission-critical applications and data in a timely and organized fashion.
Also to be considered, they said, are data currency, network bandwidth, host dependencies and impact, storage dependencies, replication configurations and scenarios, consistency technology, application failover and failback, application awareness, scale and manageability.
Beyond its main role in data recovery, replication technology can make an organization more nimble, the researchers said. For example, an IT department can develop and test applications on replicated data at the recovery site. So a credit union moving its data center can use replicated data to minimize system downtime, and content distribution can benefit from replication's capabilities.
Replication systems tend to fall into one of four categories: storage, appliance, host or database. Each has pros and cons as well as different use case scenarios, the Forrester Report said.
Storage-based replication is the granddaddy of these systems and usually the most expensive. In this approach, data is replicated between sites at the storage volume level. It has been in use for about 20 years now to replicate data synchronously, asynchronously or periodically across heterogeneous operating systems, including mainframe systems.
Products are widely available from the major vendors. The major drawback is cost: storage replication is often licensed per terabyte, with a base license fee of $10,000 to $25,000 for the first terabyte plus an additional license fee of $4,000 to $5,000 for every range of TB (2 to 6 TB, 7 to 11 TB) thereafter, the Forrester report said.
With appliance replication, an appliance continuously captures and tracks all data changes and can recreate virtualized time-stamped snapshots (known as continuous data protection, or CDP), then replicate the data to another site. The result combines traditional disaster recovery and protection against data corruption.
In addition, some replication appliances can facilitate application failover and failback. Appliance replication is both storage- and host-agnostic (except for mainframes, which it doesn't support, the Forrester report said). There are a number of possible configurations: in-band and out-of-band as well as network-based, host-based and storage array-based.
A host-based replication solution relies on an agent on the host to capture updates at the file system or volume level and direct them to an in-memory buffer pool where they are scheduled for transmission to the remote host; there, the reverse process occurs. Host-based replication is storage-agnostic. Host-based replication is licensed per host (source and target) and costs between $1,500 and $2,000 for a base license, Dines and Balaouras said. Licensing increases with advanced functionality and virtualization support.
Database replication is either embedded in the database itself, as with DB2, Oracle, Microsoft SQL Server and Sybase, or available through third-party software, the report said.
Typically, embedded database replication is included at no additional cost and tightly integrated with the database-a definite drawback, the researchers said. Even with third-party software solutions that support multiple database types, the approach is still limited to the database environment and another application is needed to protect data that is outside the database.
Effective disaster recovery isn't necessarily a single-solution strategy. "When considering which replication technologies to invest in, remember that you don't need one technology for all applications," Dines and Balaouras said.
"Without going point-product crazy, different kinds of replication can be utilized for different levels of application criticality," the Forrester analysts said. For example, using storage-based replication only for most ritical applications but host-based replication for business-critical applications.