diff --git a/presto-native-execution/presto_cpp/main/PrestoServer.cpp b/presto-native-execution/presto_cpp/main/PrestoServer.cpp
index eec870ed97394..84bff2c30e2bf 100644
--- a/presto-native-execution/presto_cpp/main/PrestoServer.cpp
+++ b/presto-native-execution/presto_cpp/main/PrestoServer.cpp
@@ -434,6 +434,9 @@ void PrestoServer::run() {
         << "': threads: " << pGlobalIOExecutor->numActiveThreads() << "/"
         << pGlobalIOExecutor->numThreads();
   }
+
+  PRESTO_SHUTDOWN_LOG(INFO) << "Release resources in AsyncDataCache";
+  cache_->prepareShutdown();
 }
 
 void PrestoServer::yieldTasks() {
@@ -492,9 +495,8 @@ void PrestoServer::initializeVeloxMemory() {
           asyncCacheSsdCheckpointGb << 30,
           asyncCacheSsdDisableFileCow);
     }
-    cache_ = std::make_shared<cache::AsyncDataCache>(
-        allocator_, memoryBytes, std::move(ssd));
-    allocator_ = cache_;
+    cache_ = cache::AsyncDataCache::create(allocator_.get(), std::move(ssd));
+    cache::AsyncDataCache::setInstance(cache_.get());
   } else {
     VELOX_CHECK_EQ(
         systemConfig->asyncCacheSsdGb(),
diff --git a/presto-native-execution/presto_cpp/main/PrestoServer.h b/presto-native-execution/presto_cpp/main/PrestoServer.h
index 0fd188ed4b5d8..d554a04a2dc2d 100644
--- a/presto-native-execution/presto_cpp/main/PrestoServer.h
+++ b/presto-native-execution/presto_cpp/main/PrestoServer.h
@@ -165,17 +165,12 @@ class PrestoServer {
   // Executor for exchange data over http.
   std::shared_ptr<folly::IOThreadPoolExecutor> exchangeExecutor_;
 
-  // If not null,  the instance of AsyncDataCache used for in-memory file cache.
-  std::shared_ptr<velox::cache::AsyncDataCache> cache_;
-
   // Instance of MemoryAllocator used for all query memory allocations.
-  //
-  // NOTE: AsyncDataCache implements MemoryAllocator interface and wraps on top
-  // of a real memory allocator for the actual memory allocation. So if 'cache_'
-  // has been set, 'allocator_' is also set to 'cache_'. It provides memory
-  // allocations for both file cache and query memory.
   std::shared_ptr<velox::memory::MemoryAllocator> allocator_;
 
+  // If not null,  the instance of AsyncDataCache used for in-memory file cache.
+  std::shared_ptr<velox::cache::AsyncDataCache> cache_;
+
   std::unique_ptr<http::HttpServer> httpServer_;
   std::unique_ptr<SignalHandler> signalHandler_;
   std::unique_ptr<Announcer> announcer_;
diff --git a/presto-native-execution/presto_cpp/main/QueryContextManager.cpp b/presto-native-execution/presto_cpp/main/QueryContextManager.cpp
index 4fc69c2db7e4b..c07e8e39d82ea 100644
--- a/presto-native-execution/presto_cpp/main/QueryContextManager.cpp
+++ b/presto-native-execution/presto_cpp/main/QueryContextManager.cpp
@@ -135,7 +135,7 @@ std::shared_ptr<core::QueryCtx> QueryContextManager::findOrCreateQueryCtx(
       executor().get(),
       std::move(configStrings),
       connectorConfigs,
-      memory::MemoryAllocator::getInstance(),
+      cache::AsyncDataCache::getInstance(),
       std::move(pool),
       spillExecutor(),
       queryId);
diff --git a/presto-native-execution/presto_cpp/main/types/PrestoToVeloxQueryPlan.cpp b/presto-native-execution/presto_cpp/main/types/PrestoToVeloxQueryPlan.cpp
index a20a3eace53b8..d02144f19e96b 100644
--- a/presto-native-execution/presto_cpp/main/types/PrestoToVeloxQueryPlan.cpp
+++ b/presto-native-execution/presto_cpp/main/types/PrestoToVeloxQueryPlan.cpp
@@ -2574,9 +2574,9 @@ core::PlanFragment VeloxQueryPlanConverterBase::toVeloxQueryPlan(
             planFragment.planNode =
                 std::make_shared<core::PartitionedOutputNode>(
                     "root",
+                    core::PartitionedOutputNode::Kind::kPartitioned,
                     partitioningKeys,
                     numPartitions,
-                    core::PartitionedOutputNode::Kind::kPartitioned,
                     partitioningScheme.replicateNullsAndAny,
                     std::make_shared<RoundRobinPartitionFunctionSpec>(),
                     outputType,
@@ -2594,9 +2594,9 @@ core::PlanFragment VeloxQueryPlanConverterBase::toVeloxQueryPlan(
             planFragment.planNode =
                 std::make_shared<core::PartitionedOutputNode>(
                     "root",
+                    core::PartitionedOutputNode::Kind::kPartitioned,
                     partitioningKeys,
                     numPartitions,
-                    core::PartitionedOutputNode::Kind::kPartitioned,
                     partitioningScheme.replicateNullsAndAny,
                     std::make_shared<HashPartitionFunctionSpec>(
                         inputType, keyChannels, constValues),
@@ -2642,9 +2642,9 @@ core::PlanFragment VeloxQueryPlanConverterBase::toVeloxQueryPlan(
 
     planFragment.planNode = std::make_shared<core::PartitionedOutputNode>(
         "root",
+        core::PartitionedOutputNode::Kind::kPartitioned,
         partitioningKeys,
         numPartitions,
-        core::PartitionedOutputNode::Kind::kPartitioned,
         partitioningScheme.replicateNullsAndAny,
         std::make_shared<HivePartitionFunctionSpec>(
             hivePartitioningHandle->bucketCount,
diff --git a/presto-native-execution/velox b/presto-native-execution/velox
index 273679c06e4df..6516bbba0e792 160000
--- a/presto-native-execution/velox
+++ b/presto-native-execution/velox
@@ -1 +1 @@
-Subproject commit 273679c06e4dfcf47c3af6270a76b2b1f38ee367
+Subproject commit 6516bbba0e7925730376197b358f5a746c8a308d