Features of a good replication software
OK, you've decided that data replication is the way to go for your organization to protect its data and virtually eliminate downtime due to hardware or software failure. What features should replication software have to provide the utmost in functionality?
- Look for byte-level dataset changes - Once the selected drives, directories and/or files have been mirrored, how data changes and is replicated can make a difference in how the network is impacted. Byte-level changes keep the consumed bandwidth low, whereas data that is replicated in large blocks or as an entire file every time it is changed can consume precious network resources.
- Cross-platform support - Support for multiple operating systems such as Windows, NetWare and Unix platforms ensures that the software will function in any heterogeneous network environment with the most commonly deployed operating systems.
- Source server automated failover - Setting a target server so that it can automatically assume the functions of the source server can either minimize or eliminate downtime. Scripting functions allow the target server to start up applications and/or services so that normal operations continue with little or no delay to end users.
- Multiple source server failover - Setting a target server so that it can automatically assume the functions of a number of source servers.
- over a WAN to enable remote replication of data for off-site disaster recovery - Asynchronous data replication technology minimizes any performance impact on the source server due to slow links or loss of remote connectivity. Byte-level replication is the best type of replication software for remote replication because it is less impacted by slow connections and doesn?t consume excessive remote network traffic.
- Supports cluster technology by replicating a cluster node to another location or off-site - to protect data against geographic or site disaster and eliminate a cluster depending on a single shared storage unit.
- Real-time and queued replication - Real-time replication means that there is always a duplicate copy of critical data but sometimes if the server is otherwise involved, or bandwidth conservation needs to be practiced at specific times, then the replicated data should be able to be queued. That is, the changes are held until a specified time, and then they are downloaded to the target server.
Not every replication software will have all of these features, and some will implement more successfully in your environment than others. Subject any replication software that you will be using to protect business critical data to a test of its ability to accurately replicate.
About the author: This tip was submitted and prepared by Dale Mescher from NSI Software, makers of data replication software Double-Take and GeoCluster based in Hoboken, NJ. To find out more about NSI Software go to their Web site at http://www.nsisoftware.com/
This was first published in February 2001