Cache RecommendationManager to improve the avg response time

taar_api/api/views.py

@@ -1,3 +1,4 @@
+from django.core.cache import cache
 from django.conf import settings
 from django.http import JsonResponse
 
@@ -8,8 +9,14 @@ from taar.hbase_client import HBaseClient
 
 def recommendations(request, client_id):
     """Return a list of recommendations provided a telemetry client_id."""
-    hbase_client = HBaseClient(settings.HBASE_HOST)
-    profile_fetcher = ProfileFetcher(hbase_client)
-    recommendation_manager = RecommendationManager(profile_fetcher)
+    recommendation_manager = cache.get("recommendation_manager")
+    if recommendation_manager is None:
+        hbase_client = HBaseClient(settings.HBASE_HOST)
+        profile_fetcher = ProfileFetcher(hbase_client)
+        recommendation_manager = RecommendationManager(profile_fetcher)
+        # Cache the recommendation manager for 1h
+        cache.set("recommendation_manager",
+                  recommendation_manager,
+                  3600)
     recommendations = recommendation_manager.recommend(client_id, settings.TAAR_MAX_RESULTS)
     return JsonResponse({"results": recommendations})