Abstract—Institutional correspondence systems require Arabic search that remains responsive at scale while enforcingaccess policies within the query execution path. This paperpresents an applied benchmark study of ICAS1, a Laravelbased institutional correspondence and archiving system backedby PostgreSQL. The evaluation uses a dedicated syntheticbenchmark database containing approximately 100,000 correspondences, 60,000 archives, and 120,000 recipient links. PostgreSQL Arabic full-text search (FTS) using to_tsvector andplainto_tsquery with Generalized Inverted Index (GIN)indexing is compared against an ILIKE baseline across threeprimary institutional scenarios. Latencies are measured usingEXPLAIN (ANALYZE, BUFFERS) with seven repeated runs.Results show FTS achieves median latencies of 21–55ms compared with 229–248ms for ILIKE, yielding speedups of 4.2×to 11.7×. An extended evaluation across 10 Arabic institutionalkeywords confirms a consistent warm-cache speedup of 4.15×4.98× for the global administrator scenario and 10.3×–12.7× fordepartment-scoped roles. Session-level cold-cache measurements(after DISCARD ALL and reconnect) show the same directionaladvantage. Retrieval overlap analysis over 30 keyword–scenariocombinations yields Jaccard = 1.0 for the corpus vocabulary used.Access-control validation using canary records confirms completeagreement between expected and observed visibility, includingnon-leakage of classified documents to restricted roles. The findings support combining Arabic FTS, query-level authorization,and GIN indexing for efficient and policy-consistent retrieval ininstitutional workloads, with the caveat that measurements reflectsingle-node conditions on a fixed synthetic corpus, and generalization to full cold-start or distributed deployments requires furtherinvestigation.