Data Protection and Information Lifecycle Management

What do you do when there is no acceptable amount of downtime? That is a problem that PdMain, Inc., of Burlington, Ontario, Canada, has struggled with. PdMain provides integrated systems that are used to manage manufacturing companies as stand-alone systems or hosted systems. They also provide hot-site capabilities for those customers that prefer to host their own applications. To provide these services, PdMain maintains its own data center near Toronto.

Customers rely on the PdMain systems and data centers for the most basic operations, including running the factory floor and dealing with the challenges of just-in-time inventory. Typical allowable downtime for PdMain customers is less than five minutes a year, and recovery from failure must happen in less than two minutes. Traditional backups, no matter how fast, cannot meet this requirement.

The ramifications of system failure in this environment are simple entire factories will shut down. Worse, the factories of the vendors feeding the factory will also shut down, because they will not know what parts need to be shipped where. The financial impact of a system failure such as this is enormous, with multiple sites and thousands of workers idled.

PdMain used all the traditional means for ensuring availability in systems. It deployed redundant, high-availability servers. Multiple telecommunications lines have been implemented, using two different carriers that use completely separate infrastructure. PdMain even maintains generators at the data center with fuel on hand for two days of operation. Unfortunately, this was not good enough. Even if the customer could cut over to a new set of servers and storage immediately, the data was old. This led to lost data and poor referential integrity. Simply put, the systems would not run without current data.

The solution was to provide continuous replication to a server at a hot site. Partnering with XOSoft, Inc., a maker of data replication software products, PdMain is able to replicate, in real time, all the databases that are the underpinnings of the application. All the data in the duplicate system is exactly as it is in the main system. The actual replication happens in milliseconds. In some cases, PdMain customers will bunker data, keeping a duplicate of the data on storage within their data centers and replicating it further to the PdMain data center. In this way, they protect themselves against local system failure and data center disasters. The use of asynchronous replication allows PdMain and its customers to use less expensive telecommunications, as low as a T1 or a DSL line, while maintaining high service levels.

By replicating data, PdMain customers are capable of continuous operation in the face of a variety of disasters. Replication is a necessary part of their business operations and the key to avoiding crippling data loss.