Softek Replicator delivers platform-friendly backup
Solution enables TCP/IP replication with any OS, file system, or hardware
Data replication is useful in so many ways. In addition to its critical role in disaster recovery, it can consolidate data from multiple servers or data stores, or distribute data to multiple stores, allowing branch offices to access local copies of data and thus reduce WAN traffic.
Although many data storage vendors provide replication functions, they typically work only with that vendor’s products, often require extra components to replicate over a TCP/IP network, and may only work between specific models of storage devices. Fibre channel-based systems, such as Maxxan’s MXV320 Intelligent Application Switch, can provide transparent data replication and work between different types of systems, but they are relatively expensive, starting at about $30,000.
Fujitsu Softek’s server-based replication software, Replicator 2.1, supports a wide variety of Unix platforms, Linux, and Windows, with a very low entry price of $2,300. It supports replication over TCP/IP without additional hardware and imposes a minimal load on the server performing the replication. Intended primarily for one-to-one replication for disaster recovery, Replicator also provides simple one-to-many or many-to-one replication, and it offers checkpoint capabilities for backups of databases and other software that typically keeps files open.
Replicator works at the block level, so it is transparent to the OS and can set replication up between any two systems that use the same type of file system, be it hardware or server OS. The partitions should be the same size, but even that is not a strict requirement; the primary partition can be smaller than the secondary.
Creating a replicated partition is simple. First, you install Replicator on the primary and secondary servers, a five-minute process. You can conveniently administer Replicator across your enterprise from a single, well-designed GUI console. You can configure and update it via the command line or scripts. The software is easy to use and has powerful command-line utilities.
Once the software is installed, you designate a partition on each system, then initiate replication. If there is a lot of data on the primary partition, you may prepopulate the secondary partition from backup tape or optical media, then synchronize the partitions over the network, incorporating only the changes made since that backup. This reduces network traffic.
Once replication is established, data is duplicated with the original write-ordering and with full tracking of changes. Replicator uses journaling at both ends of the replication pair to ensure data integrity. You may also pause replication by creating a checkpoint. This causes the data log on the primary volume to store all changes but not pass them on to the secondary volume, thus allowing a backup to be made from the secondary system without losing availability of data.
If you desire, you can create a third link in a chain: a primary partition replicating to a secondary partition, the latter of which replicates to a third backup partition. This lets you create a checkpoint between the second and third partitions, preserving replication for disaster recovery while still allowing for a snapshot for backup purposes.
Replicator can also replicate one primary volume to several secondary volumes to allow local access to data, and it can replicate several primary volumes to one secondary volume, allowing server or data consolidation without loss of access to the data at any time.
In my test, Replicator produced minimal overhead, adding about 2 percent utilization to a Windows server under load when replication was enabled. Replicator operates with all disk subsystems supported by AIX, HP-UX, Solaris, and Windows NT and 2000. It installs as a service or daemon and is compatible with nearly all common file systems, volume managers, database management systems, and disk utilities.
Any admin looking for a simple and inexpensive way to set up replication of data between different hardware systems over TCP/IP should investigate Replicator 2.1.