What are the implications of DRIVER=SQL Anywhere Native?

[Breck_Carter]

The special "DRIVER=SQL Anywhere Native" syntax is briefly described in the CREATE SERVER Help topic.

What are the advantages to using this syntax? Performance?

Is it OK to use it for a SQL Anywhere 16 database that is creating remote server connections to earlier versions of SQL Anywhere? E.g. V16 connecting to V12, to V11, to V10, to ... ?

How far back is OK?

[Former Member]

It also has a mention in the SQL Anywhere New Features article as well.

This feature permits you to use our driver even when the driver is not registered. It loads the driver version that matches the database server version. It is not required to have DSN-less connections {since ODBC Driver Managers normally can facilitate that with the ServerName='Driver=...' usage}.

Some performance differences come about by not doing the DRIVER and DSN lookups during the connection phase. Further each ODBC SQL*() function call has a little less overhead since it is not passing through the Platform/3rd party(on Unix/Linux) driver manager.

It will also get some advantages of loading the Driver version matched to the server version. And, with that, backward compatibility with other (older) server versions should carry the day their as well; something we've usually done quite well if I may say so. So that part is okay too.

I won't use the "varying mileage" parable to describe the degree of benefit but I do suspect performance differences to only be modest in most instances.

I'm sure you will let us know if you discover any significant difference or instabilities but I suspect this to be mostly transparent.

The main caveat here is that an ODBC Driver Manager will used under the documented restrictions so the above will not apply in that case.

HTH