Actually I just attended a Shareplex session at IOUG in Orlando last week and here is what I know. It is similar to a standby database in that it is a form of data replication (as opposed to storage replication like Veritas) but its not the same. It reads the redo logs at the O/S level and translates them to SQL statements then transfers these SQL statements to the target database through TCP/IP (but does not use Sql*Net). So in other words there is no real overhead on the source database because it does not use the source db engine to read the redo logs. It only connects to the source db briefly to read the data dict for object names using the object id from redo. The target database is available throughout this replication process, unless a standby database that is not accessible while in recovery mode. It can also help with backups in that you can shutdown the target database and do a cold backup of it leaving the source db available. Shareplex contends that it can keep up with 3.5G of redo generation an hour and 55G of redo generation a day. Also, Shareplex does support some DDL statements such as truncates and column changes, whereas Standby does not, and it can delay the application of the SQL statements to the target. Shareplex is command line driven and easily scriptable. And it can handle replication between different Oracle versions, nice if you want to upgrade Oracle versions using production data (aka target db) without risking it on the real production database first. And lastly, if there is disk block corruption in the redo log, it is not propogated to the target database.

I was pretty impressed by the session, however, I hear Shareplex is kind of pricey. We are thinking about going to Shareplex. If anybody else out there uses Shareplex please let us know if it is a good High Availibity option, or if there are any pitfalls to it.

Thanks.