Why does a scrolling query generate more database reads in Client-Server mode?

Why does a scrolling query generate more database reads in Client-Server mode?

Why does a scrolling query generate more database reads in Client-Server mode than in Self-Service mode?

Executing a variation of the following query:

DEFINE QUERY myquery FOR mytable SCROLLING.
OPEN QUERY myquery FOR EACH mytable NO-LOCK.
GET FIRST myquery.

Executing the same query with a Client-Server connection generates more reads than executing it with a self-service connection.

The number of reads become the same after removing either "SCROLLING" OR "NO-LOCK".

This is expected behavior, caused by the use of the SCROLLING option. Per the Online Help,

"The SCROLLING option can improve NETWORK performance because query NO-LOCK records can be prefetched with multiple records per network message."

More reads are generated with the client-server connection because the AVM is prefetching table records in order to improve performance for forward or backward movement within the query's result list.