I have been studying a document I downloaded from Oracle, entitled "SQL*Net
Performance tuning Utilizing Underlying Network Protocol." At first I thought
it might be quite useful, but the more I study it, the more I begin to wonder .
. . the author made several assertions that left me scratching my head.
At one point, speaking of the TNSNAMES.ORA file, he says, "Instead of reading
the entire file and scanning for the relevant entry, it is better to implement
an indexing method." Later, under "Recommendations/Standards" he lists "3.
implementing indexing method to read Tnsnames.ora file"
Now, I've never heard or read of a mechanism for indexing access to TNSNAMES.
So who is this recommendation for? Certainly we DBAs and developers have no way
of doing this, so is he directing this at an internal Oracle audience
responsible for SQL*Net implementation? Or is he just being uselessly
theoretical?
He also spends a lot of time talking about configuring Array Size, and in his
"Recommendations/Standards" he lists "2. application must have the capability to
determine the Array Size." Again, nice theory but unless your apps run under
SQL*Plus (the client he used for his testing) of what value is it? My days as a
developer were spent mostly with COBOL so I don't have much hands-on experience
with the development tools my databases are supporting. Do Powerbuilder and VB
with ADO have methods of controlling what Plus calls "array size"?
On adjusting network buffer sizes he says the following:
"SQL*NET should be intelligent enough to adjust its buffer size automatically
according to the UNP (underlying network protocol) buffer size."
Well, perhaps it SHOULD be, but if it isn't, then it's a moot point.
Or am I missing something?
--
Ed Stevens
(Opinions expressed do not necessarily represent those of my employer.)