From 561c87222df972748f3f6bc884078eee99db76ad Mon Sep 17 00:00:00 2001
From: taiyang-li <654010905@qq.com>
Date: Thu, 26 May 2022 11:04:35 +0800
Subject: [PATCH] add prefetch for hive text

---
 src/Storages/HDFS/AsynchronousReadBufferFromHDFS.cpp | 7 +++++++
 src/Storages/HDFS/AsynchronousReadBufferFromHDFS.h   | 1 +
 src/Storages/Hive/StorageHive.cpp                    | 3 +++
 3 files changed, 11 insertions(+)

diff --git a/src/Storages/HDFS/AsynchronousReadBufferFromHDFS.cpp b/src/Storages/HDFS/AsynchronousReadBufferFromHDFS.cpp
index c18364306d0..35dccf644d6 100644
--- a/src/Storages/HDFS/AsynchronousReadBufferFromHDFS.cpp
+++ b/src/Storages/HDFS/AsynchronousReadBufferFromHDFS.cpp
@@ -42,9 +42,12 @@ AsynchronousReadBufferFromHDFS::AsynchronousReadBufferFromHDFS(
     , impl(std::move(impl_))
     , prefetch_buffer(settings_.remote_fs_buffer_size)
     , read_until_position(impl->getFileSize())
+    , use_prefetch(settings_.remote_fs_prefetch)
     , log(&Poco::Logger::get("AsynchronousReadBufferFromHDFS"))
 {
     ProfileEvents::increment(ProfileEvents::RemoteFSBuffers);
+    if (use_prefetch)
+        prefetch();
 }
 
 bool AsynchronousReadBufferFromHDFS::hasPendingDataToRead()
@@ -142,6 +145,10 @@ bool AsynchronousReadBufferFromHDFS::nextImpl()
 
     file_offset_of_buffer_end = impl->getFileOffsetOfBufferEnd();
     prefetch_future = {};
+
+    if (use_prefetch)
+        prefetch();
+
     return size;
 }
 
diff --git a/src/Storages/HDFS/AsynchronousReadBufferFromHDFS.h b/src/Storages/HDFS/AsynchronousReadBufferFromHDFS.h
index 43ab0c09834..dd86ffac1d4 100644
--- a/src/Storages/HDFS/AsynchronousReadBufferFromHDFS.h
+++ b/src/Storages/HDFS/AsynchronousReadBufferFromHDFS.h
@@ -57,6 +57,7 @@ private:
 
     size_t file_offset_of_buffer_end = 0;
     std::optional<size_t> read_until_position;
+    bool use_prefetch;
 
     Poco::Logger * log;
 };
diff --git a/src/Storages/Hive/StorageHive.cpp b/src/Storages/Hive/StorageHive.cpp
index 6a8ea4b4ca5..0f21403673c 100644
--- a/src/Storages/Hive/StorageHive.cpp
+++ b/src/Storages/Hive/StorageHive.cpp
@@ -133,6 +133,9 @@ public:
             if (to_read_block.has(name_type.name))
                 to_read_block.erase(name_type.name);
         }
+
+        /// Apply read buffer prefetch for HiveText format, because it is read sequentially
+        read_settings.remote_fs_prefetch = format == "HiveText";
     }
 
     FormatSettings updateFormatSettings(const HiveFilePtr & hive_file)