Replication, Replication, Replication.

Data replication software's role in the marketplace continues to expand

Data replication has historically applied to databases, not only to copying, but also to synchronizing a set of replicas so that changes made to one are reflected in all. The basic idea of replication is to set up a program to scan at intervals for changes to specified files, at which time it replicates the changes to a storage target. Unlike mirroring, the replication only occurs with specified databases, logical volumes or file systems, and can handle a many-to-one operation between multiple production servers and the storage site. Depending on the software or user parameters, it may replicate whole files or only changed bytes. Data replication software is expanding in the marketplace.

The market is growing, growing fast. "The threat of lost economic viability in the event of a disaster will drive penetration of data replication technologies in large enterprises from less than 25% to 75% by 2004," says Donna Scott, vice president of research and advisory services of the Gartner Group.

Driving factors include:

1. Millions of PC users are doing complicated analyses, knowledge management, and sophisticated database queries from their own computers. Join that to ever-increasing data production and storage needs and you have a hugely expanded user base with a lot of disconnected information. This information is at risk from several directions, including failure to back up, failure to share information, and failure to update data for other users.

2. Dispersed data has also paved the way for replication software. With many companies growing from one-site operations to expanded organizations, it's harder and harder to manage data collections for business intelligence, as well as ensure backup at multiple remote Sites.

3. With a return to server-centric computing, companies are depending more and more on enterprise data management and centralization. However, since individual access to computing power and data production remains a reality, there's no return to a largely mainframe environment. Data replication can provide the bridge between client-and server-centric models.

Data replication, however, is a growing challenge. Working to achieve higher levels of data availability, system administrators increasingly create multiple copies of business-critical data to quickly recover from disasters, load-balance access to data across Internet servers, and offload primary servers from operations such as backup and recovery or data mining. As data centers attempt to maintain data availability in the event of local catastrophes and globally service customers, multiple copies of data must also be efficiently distributed and synchronized around the world. Traditional methods of backup and recovery to tape are sometimes used, but have long recovery times and are not effective for real-time distribution of data. Other alternatives such as array replication allow two remote arrays to be connected and synchronized, but these solutions are specific to each manufacturer and require special infrastructures to run over long distances.

One solution is a data replication scheme using host software such as VERITAS Software's new data replication family, which replicates data between one or more servers over a standard IP network. These solutions enable system administrators to select the storage-array vendor at each site, allowing the customer to optimize costs of secondary site equipment. In addition, since secondary copies are kept up-to-date on online storage, the time to restore data availability after a disaster or major system outage is significantly reduced. Host data replication solutions can be used to propagate changes at the database, logical volume, or file system level. The result is an efficient and flexible solution for the business continuity of global operations.

According to Informix, a successful replication system must do three things: 1) give high performance without overburdening the source database or requiring new programming, 2) provide high availability and consistent data delivery, and 3) allow database administrators to configure and manage the replication system efficiently and effectively.

DataMirror adds some advanced features to look for in a replication software.

* Selectivity -- For data warehouses, can the software selectively filter data sets by row and/or record?

* Data Transformation -- For data warehouses, can the software recognize dissimilar coding or obscure naming conventions and consolidate the information or distribute the data to remote locations?

* Support for Heterogeneous Environments -- Can the software operate in mixed system computing environments?

* Bi-directional Replication -- Does the software allow users to freely move data between systems, enabling true data sharing applications?

With the enormously expanding storage and database needs of today's business, it's increasingly important to protect and manage the databases that are the lifeblood of business decisions. Combined with the reality of growing IT responsibilities with static budget and staffing, data replication presents a tremendous tool for marketing business continuity, data sharing, and synchronization solutions. As mission-critical applications and data proliferate throughout the enterprise and as IT departments are charged with managing and protecting this data across distance, mixed platforms, and data transport systems, data replication products that meet these requirements fill a distinct need in the marketplace.