Like so many other good technical questions, the answer to this one is "it depends."
It depends on how often your data is updated and how well those updates can be shared among the file servers. With clustered file servers, your disks, and therefore your data, are shared between the servers, and if one server's data gets updated, the other server's gets updated at the same time. With the network load balanced approach, data must be replicated between multiple servers. If your data changes a lot, then this adds some serious complexity as you attempt to keep the data in sync. If you lose a server while the data is out of sync, your users will likely see corrupted data. Is the process that keeps the servers in sync a single point of failure? What happens if it fails? Network load balancing is an excellent approach if data is relatively static, and is commonly used for read-mostly Web servers.
The other factor to consider is load. Can a single server handle the load that will be placed on it? If you need to distribute the load among several machines in parallel, then clustering will probably not do the job, and you'll need to look at network load balancing.
I hope this helps.
Evan L. Marcus
Editor's note: Do you agree with this expert's response? If you have more to share, post it in one of our .bphAaR2qhqA^0@/searchstorage>discussion forums.
This was first published in July 2002