Shareplex & Veritas Quick IO (18776)

Both versions of Veritas quick I/O (cache and nocache) are supported, if cached you will have to enable SP_OCT_REDOLOG_ENSURE so we wait 2 seconds to make sure that the block is completely flushed by Veritas.

if SP_OCT_REDOLOG_ENSURE is not set and veritas quick I/O with caching is enabled we might not read the correct information from the redologs.

Without cache, quick I/O is a regular file system, the following is an extract from the Shareplex Admin Manual 5.3, which explains this:

Because of the way Quick I/O caches disk information, Capture cannot access the cache, so it interprets that it already processed the data and waits for more to process. While it waits, it loses pace with Oracle, which continues generating redo data not accessible to Capture. The eventual result is a log wrap and the necessity to access the archive logs. When Capture processes what it missed and returns the redo logs, the same thing happens again because of the Quick I/O cache. The only workaround is to remove Quick I/O from the redo and archive logs, but you can use it for the datafiles.