What are your views about DR focused software products (i.e. Veritas, Legato/Octopus or maybe DoubleTake) vs. hardware-based replication (i.e. controller-based on EVA or Clariion)? I presume that the software solution is more prone to error and therefore less reliable due to the nature of software and human error but I am very interested in your point of view.
I'm not sure why you'd conclude that software products are more prone to human error than hardware products. It's all software, whether the software is written by Veritas (my employer), EMC, Legato, EVA or someone else, it's all software. The real difference is in where the software runs. Does it run on the server or on a disk array? And, all software is prone to bugs, and most software configurations have complexities.
Both hardware- and software-based solutions protect data and generally do so equally well. Hardware replication can send data between any two devices that can speak to the disk array, while software replication allows data to be sent between just about any type of storage, even if mixed vendors are involved.
I have already written both answers in this space and columns about synchronous vs. asynchronous replication, the trade-off between performance and data currency, and how not all hardware replication can do asynchronous replication while maintaining data integrity. I have also written about my experiences with replication performance and how, contrary to what you might think, software replication tends to outperform hardware replication in a fair fight.
Evan L. Marcus
Editor's note: Do you agree with this expert's response? If you have more to share, post a comment below.
This was first published in January 2003