Objective: Calculate a baseline TCP dispatcher count, specify a per-dispatcher connection limit, and distinguish those settings from shared-server process sizing.
The host dilbert runs a single-instance Oracle AI Database 26ai CDB. An open PDB provides the existing, started application service fred. A TCP listener already accepts connections on port 1521, and service registration is configured. Shared-server use has not been disabled for the PDB.
The application has an average of 150 local and 50 remote concurrent client connections. Both groups now use TCP; the legacy SPX configuration is outside this exercise. Treat the combined 200 connections as the baseline to size, not as a measured peak. The lab policy permits at most 50 network connections per dispatcher. Assume no Oracle Connection Manager session multiplexing and no other application connections using this dispatcher pool.
For a repeatable lab answer, start with 5 shared-server processes and allow growth to 20. These are supplied trial values, not numbers derived from the connection count. Assume that CPU, memory, and the instance process budget have been reviewed for this lab. Production sizing requires representative workload measurements.
Complete this PFILE-style fragment as a proposed instance configuration. It is not a SQL script and does not belong in listener.ora. In a CDB, configure the instance dispatcher and shared-server pool settings from the root. Existing dispatcher configurations must be reviewed before applying a change.
# Proposed instance settings for the dilbert training host
DISPATCHERS = '(PROTOCOL=TCP)(DISPATCHERS=__)(CONNECTIONS=__)'
SHARED_SERVERS = __
MAX_SHARED_SERVERS = __
The existing service and listener are prerequisites; you do not need to create another listener for remote clients. To request shared-server handling, include (SERVER=SHARED) in the client's CONNECT_DATA along with (SERVICE_NAME=fred).
Each task is worth 2 points. The results page provides a matching model answer and rubric. This free-text exercise does not calculate an automatic score. No live database access is required; describe expected verification results if you cannot run a lab.