From 97b923fd7be36d4b0bcee98c45dc7c18dfc6e199 Mon Sep 17 00:00:00 2001
From: Alexey Milovidov <milovidov@yandex-team.ru>
Date: Sun, 8 Sep 2013 07:30:52 +0000
Subject: [PATCH] dbms: uncompressed cache: better [#CONV-8661].

---
 dbms/include/DB/Storages/StorageMergeTree.h |  9 ++++++++-
 dbms/src/Storages/StorageMergeTree.cpp      | 12 ++++++++++++
 2 files changed, 20 insertions(+), 1 deletion(-)

diff --git a/dbms/include/DB/Storages/StorageMergeTree.h b/dbms/include/DB/Storages/StorageMergeTree.h
index c894b54280a..35b4266afe6 100644
--- a/dbms/include/DB/Storages/StorageMergeTree.h
+++ b/dbms/include/DB/Storages/StorageMergeTree.h
@@ -67,6 +67,11 @@ struct StorageMergeTreeSettings
 	
 	/// Если отрезок индекса может содержать нужные ключи, делим его на столько частей и рекурсивно проверяем их.
 	size_t coarse_index_granularity;
+
+	/** Максимальное количество строк на запрос, для использования кэша разжатых данных. Если запрос большой - кэш не используется.
+	  * (Чтобы большие запросы не вымывали кэш.)
+	  */
+	size_t max_rows_to_use_cache;
 	
 	StorageMergeTreeSettings() :
 		max_size_ratio_to_merge_parts(5),
@@ -75,7 +80,8 @@ struct StorageMergeTreeSettings
 		merging_threads(2),
 		min_rows_for_concurrent_read(20 * 8192),
 		min_rows_for_seek(5 * 8192),
-		coarse_index_granularity(8) {}
+		coarse_index_granularity(8),
+		max_rows_to_use_cache(1024 * 1024) {}
 };
 
 /// Пара засечек, определяющая диапазон строк в куске. Именно, диапазон имеет вид [begin * index_granularity, end * index_granularity).
@@ -171,6 +177,7 @@ private:
 	
 	size_t min_marks_for_seek;
 	size_t min_marks_for_concurrent_read;
+	size_t max_marks_to_use_cache;
 
 	/// Для схлопывания записей об изменениях, если это требуется.
 	String sign_column;
diff --git a/dbms/src/Storages/StorageMergeTree.cpp b/dbms/src/Storages/StorageMergeTree.cpp
index bf74683a78d..41d04c6d0b1 100644
--- a/dbms/src/Storages/StorageMergeTree.cpp
+++ b/dbms/src/Storages/StorageMergeTree.cpp
@@ -74,6 +74,7 @@ StorageMergeTree::StorageMergeTree(
 {
 	min_marks_for_seek = (settings.min_rows_for_seek + index_granularity - 1) / index_granularity;
 	min_marks_for_concurrent_read = (settings.min_rows_for_concurrent_read + index_granularity - 1) / index_granularity;
+	max_marks_to_use_cache = (settings.max_rows_to_use_cache + index_granularity - 1) / index_granularity;
 
 	/// создаём директорию, если её нет
 	Poco::File(full_path).createDirectories();
@@ -314,6 +315,9 @@ BlockInputStreams StorageMergeTree::spreadMarkRangesAmongThreads(
 		}
 		sum_marks += sum_marks_in_parts[i];
 	}
+
+	if (sum_marks > max_marks_to_use_cache)
+		use_uncompressed_cache = false;
 	
 	BlockInputStreams res;
 	
@@ -400,6 +404,14 @@ BlockInputStreams StorageMergeTree::spreadMarkRangesAmongThreads(
 BlockInputStreams StorageMergeTree::spreadMarkRangesAmongThreadsFinal(
 	RangesInDataParts parts, size_t threads, const Names & column_names, size_t max_block_size, bool use_uncompressed_cache)
 {
+	size_t sum_marks = 0;
+	for (size_t i = 0; i < parts.size(); ++i)
+		for (size_t j = 0; j < parts[i].ranges.size(); ++j)
+			sum_marks += parts[i].ranges[j].end - parts[i].ranges[j].begin;
+
+	if (sum_marks > max_marks_to_use_cache)
+		use_uncompressed_cache = false;
+	
 	ExpressionActionsPtr sign_filter_expression;
 	String sign_filter_column;
 	createPositiveSignCondition(sign_filter_expression, sign_filter_column);