diff --git a/dbms/include/DB/AggregateFunctions/AggregateFunctionGroupArray.h b/dbms/include/DB/AggregateFunctions/AggregateFunctionGroupArray.h
index bc8d0fc5184..fb935a782a0 100644
--- a/dbms/include/DB/AggregateFunctions/AggregateFunctionGroupArray.h
+++ b/dbms/include/DB/AggregateFunctions/AggregateFunctionGroupArray.h
@@ -15,7 +15,7 @@ namespace DB
 
 struct AggregateFunctionGroupArrayData
 {
-	Array value;
+	Array value;	/// TODO Добавить MemoryTracker
 };
 	
 
diff --git a/dbms/include/DB/AggregateFunctions/AggregateFunctionQuantile.h b/dbms/include/DB/AggregateFunctions/AggregateFunctionQuantile.h
index 1c12f2d48bb..e6b3c09327c 100644
--- a/dbms/include/DB/AggregateFunctions/AggregateFunctionQuantile.h
+++ b/dbms/include/DB/AggregateFunctions/AggregateFunctionQuantile.h
@@ -20,7 +20,7 @@ template <typename ArgumentFieldType>
 struct AggregateFunctionQuantileData
 {
 	typedef ReservoirSampler<ArgumentFieldType, ReservoirSamplerOnEmpty::RETURN_NAN_OR_ZERO> Sample;
-	Sample sample;
+	Sample sample;	/// TODO Добавить MemoryTracker
 };
 
 
diff --git a/dbms/include/DB/AggregateFunctions/AggregateFunctionQuantileTiming.h b/dbms/include/DB/AggregateFunctions/AggregateFunctionQuantileTiming.h
index 86b8cd78389..024cc11b078 100644
--- a/dbms/include/DB/AggregateFunctions/AggregateFunctionQuantileTiming.h
+++ b/dbms/include/DB/AggregateFunctions/AggregateFunctionQuantileTiming.h
@@ -2,6 +2,8 @@
 
 #include <limits>
 
+#include <DB/Common/MemoryTracker.h>
+
 #include <DB/IO/WriteHelpers.h>
 #include <DB/IO/ReadHelpers.h>
 
@@ -324,6 +326,9 @@ private:
 
 	void toLarge()
 	{
+		if (current_memory_tracker)
+			current_memory_tracker->alloc(sizeof(detail::QuantileTimingLarge));
+
 		/// На время копирования данных из tiny, устанавливать значение large ещё нельзя (иначе оно перезатрёт часть данных).
 		detail::QuantileTimingLarge * tmp_large = new detail::QuantileTimingLarge;
 
@@ -343,7 +348,12 @@ public:
 	~QuantileTiming()
 	{
 		if (isLarge())
+		{
 			delete large;
+
+			if (current_memory_tracker)
+				current_memory_tracker->free(sizeof(detail::QuantileTimingLarge));
+		}
 	}
 
 	void insert(UInt64 x)
@@ -405,6 +415,10 @@ public:
 			if (!isLarge())
 			{
 				tiny.count = TINY_MAX_ELEMS + 1;
+
+				if (current_memory_tracker)
+					current_memory_tracker->alloc(sizeof(detail::QuantileTimingLarge));
+
 				large = new detail::QuantileTimingLarge;
 			}
 
@@ -424,6 +438,10 @@ public:
 			if (!isLarge())
 			{
 				tiny.count = TINY_MAX_ELEMS + 1;
+
+				if (current_memory_tracker)
+					current_memory_tracker->alloc(sizeof(detail::QuantileTimingLarge));
+
 				large = new detail::QuantileTimingLarge;
 			}
 			
diff --git a/dbms/include/DB/Common/Arena.h b/dbms/include/DB/Common/Arena.h
index 42af26f3baa..e81c7ad9c74 100644
--- a/dbms/include/DB/Common/Arena.h
+++ b/dbms/include/DB/Common/Arena.h
@@ -6,6 +6,7 @@
 #include <Poco/SharedPtr.h>
 #include <Yandex/likely.h>
 #include <DB/Common/ProfileEvents.h>
+#include <DB/Common/MemoryTracker.h>
 
 
 namespace DB
@@ -37,6 +38,9 @@ private:
 			ProfileEvents::increment(ProfileEvents::ArenaAllocChunks);
 			ProfileEvents::increment(ProfileEvents::ArenaAllocBytes, size_);
 
+			if (current_memory_tracker)
+				current_memory_tracker->alloc(size_);
+
 			begin = allocate(size_);
 			pos = begin;
 			end = begin + size_;
@@ -47,6 +51,9 @@ private:
 		{
 			deallocate(begin, size());
 
+			if (current_memory_tracker)
+				current_memory_tracker->free(size());
+
 			if (prev)
 				delete prev;
 		}
diff --git a/dbms/include/DB/Common/HashTable/HashTable.h b/dbms/include/DB/Common/HashTable/HashTable.h
index 12c7453d1bd..231a0a14ff6 100644
--- a/dbms/include/DB/Common/HashTable/HashTable.h
+++ b/dbms/include/DB/Common/HashTable/HashTable.h
@@ -264,6 +264,7 @@ protected:
 
 	void alloc()
 	{
+		// TODO Если здесь исключение, то free будет вызывана от неправильного размера.
 		buf = reinterpret_cast<Cell *>(Allocator::alloc(bufSizeBytes()));
 	}
 
diff --git a/dbms/include/DB/Common/HashTable/HashTableAllocator.h b/dbms/include/DB/Common/HashTable/HashTableAllocator.h
index 9d7cfcce1b8..3906c351c03 100644
--- a/dbms/include/DB/Common/HashTable/HashTableAllocator.h
+++ b/dbms/include/DB/Common/HashTable/HashTableAllocator.h
@@ -4,6 +4,7 @@
 #include <string.h>
 #include <sys/mman.h>
 
+#include <DB/Common/MemoryTracker.h>
 #include <DB/Core/Exception.h>
 #include <DB/Core/ErrorCodes.h>
 
@@ -31,6 +32,9 @@ public:
 	/// Выделить кусок памяти и заполнить его нулями.
 	void * alloc(size_t size)
 	{
+		if (current_memory_tracker)
+			current_memory_tracker->alloc(size);
+
 		void * buf;
 
 		if (size >= MMAP_THRESHOLD)
@@ -63,6 +67,9 @@ public:
 		{
 			::free(buf);
 		}
+
+		if (current_memory_tracker)
+			current_memory_tracker->free(size);
 	}
 
 	/** Увеличить размер куска памяти.
@@ -72,6 +79,9 @@ public:
 	  */
 	void * realloc(void * buf, size_t old_size, size_t new_size)
 	{
+		if (current_memory_tracker)
+			current_memory_tracker->realloc(old_size, new_size);
+
 		if (old_size < MMAP_THRESHOLD && new_size < MMAP_THRESHOLD)
 		{
 			buf = ::realloc(buf, new_size);
@@ -96,7 +106,6 @@ public:
 			buf = new_buf;
 		}
 
-
 		return buf;
 	}
 };
diff --git a/dbms/include/DB/Common/MemoryTracker.h b/dbms/include/DB/Common/MemoryTracker.h
new file mode 100644
index 00000000000..85e30472899
--- /dev/null
+++ b/dbms/include/DB/Common/MemoryTracker.h
@@ -0,0 +1,49 @@
+#pragma once
+
+#include <Yandex/Common.h>
+
+
+/** Отслеживает потребление памяти.
+  * Кидает исключение, если оно стало бы больше некоторого предельного значения.
+  * Один объект может использоваться одновременно в разных потоках.
+  */
+class MemoryTracker
+{
+	Int32 amount = 0;
+	Int32 peak = 0;
+	Int32 limit = 0;
+
+public:
+	MemoryTracker(Int32 limit_) : limit(limit_) {}
+
+	~MemoryTracker();
+
+	/** Вызывайте эти функции перед соответствующими операциями с памятью.
+	  */
+	void alloc(Int32 size);
+
+	void realloc(Int32 old_size, Int32 new_size)
+	{
+		alloc(new_size - old_size);
+	}
+
+	/** А эту функцию имеет смысл вызывать после освобождения памяти.
+	  */
+	void free(Int32 size)
+	{
+		__sync_sub_and_fetch(&amount, size);
+	}
+
+	Int32 get() const
+	{
+		return amount;
+	}
+};
+
+
+/** Объект MemoryTracker довольно трудно протащить во все места, где выделяются существенные объёмы памяти.
+  * Поэтому, используется thread-local указатель на используемый MemoryTracker или nullptr, если его не нужно использовать.
+  * Этот указатель выставляется, когда в данном потоке следует отслеживать потребление памяти.
+  * Таким образом, его нужно всего-лишь протащить во все потоки, в которых обрабатывается один запрос.
+  */
+extern __thread MemoryTracker * current_memory_tracker;
diff --git a/dbms/include/DB/Common/PODArray.h b/dbms/include/DB/Common/PODArray.h
index fc6844b29d6..62e7463f7e9 100644
--- a/dbms/include/DB/Common/PODArray.h
+++ b/dbms/include/DB/Common/PODArray.h
@@ -12,6 +12,7 @@
 #include <Yandex/likely.h>
 #include <Yandex/strong_typedef.h>
 
+#include <DB/Common/MemoryTracker.h>
 #include <DB/Core/Exception.h>
 #include <DB/Core/ErrorCodes.h>
 
@@ -88,6 +89,10 @@ private:
 		}
 		
 		size_t bytes_to_alloc = to_size(n);
+
+		if (current_memory_tracker)
+			current_memory_tracker->alloc(bytes_to_alloc);
+
 		c_start = c_end = Allocator::allocate(bytes_to_alloc);
 		c_end_of_storage = c_start + bytes_to_alloc;
 	}
@@ -101,6 +106,9 @@ private:
 			::free(c_start);
 		else
 			Allocator::deallocate(c_start, storage_size());
+
+		if (current_memory_tracker)
+			current_memory_tracker->free(storage_size());
 	}
 
 	void realloc(size_t n)
@@ -117,6 +125,9 @@ private:
 		char * old_c_start = c_start;
 		char * old_c_end_of_storage = c_end_of_storage;
 
+		if (current_memory_tracker)
+			current_memory_tracker->realloc(storage_size(), bytes_to_alloc);
+
 		if (use_libc_realloc)
 		{
 			c_start = reinterpret_cast<char *>(::realloc(c_start, bytes_to_alloc));
diff --git a/dbms/include/DB/Core/ErrorCodes.h b/dbms/include/DB/Core/ErrorCodes.h
index c4f0d3f81ef..935689fd402 100644
--- a/dbms/include/DB/Core/ErrorCodes.h
+++ b/dbms/include/DB/Core/ErrorCodes.h
@@ -248,6 +248,7 @@ namespace ErrorCodes
 		FORMAT_VERSION_TOO_OLD,
 		CANNOT_MUNMAP,
 		CANNOT_MREMAP,
+		MEMORY_LIMIT_EXCEEDED,
 
 		POCO_EXCEPTION = 1000,
 		STD_EXCEPTION,
diff --git a/dbms/include/DB/DataStreams/AsynchronousBlockInputStream.h b/dbms/include/DB/DataStreams/AsynchronousBlockInputStream.h
index c5129bd18bb..acf8d8c3f36 100644
--- a/dbms/include/DB/DataStreams/AsynchronousBlockInputStream.h
+++ b/dbms/include/DB/DataStreams/AsynchronousBlockInputStream.h
@@ -89,7 +89,7 @@ protected:
 		/// Если вычислений ещё не было - вычислим первый блок синхронно
 		if (!started)
 		{
-			calculate();
+			calculate(current_memory_tracker);
 			started = true;
 		}
 		else	/// Если вычисления уже идут - подождём результата
@@ -113,13 +113,15 @@ protected:
 	void next()
 	{
 		ready.reset();
-		pool.schedule(boost::bind(&AsynchronousBlockInputStream::calculate, this));
+		pool.schedule(boost::bind(&AsynchronousBlockInputStream::calculate, this, current_memory_tracker));
 	}
 	
 
 	/// Вычисления, которые могут выполняться в отдельном потоке
-	void calculate()
+	void calculate(MemoryTracker * memory_tracker)
 	{
+		current_memory_tracker = memory_tracker;
+
 		try
 		{
 			block = children.back()->read();
diff --git a/dbms/include/DB/DataStreams/ParallelAggregatingBlockInputStream.h b/dbms/include/DB/DataStreams/ParallelAggregatingBlockInputStream.h
index 804ddfd04c7..239af6a3b16 100644
--- a/dbms/include/DB/DataStreams/ParallelAggregatingBlockInputStream.h
+++ b/dbms/include/DB/DataStreams/ParallelAggregatingBlockInputStream.h
@@ -76,7 +76,8 @@ protected:
 		for (size_t i = 0, size = many_data.size(); i < size; ++i)
 		{
 			many_data[i] = new AggregatedDataVariants;
-			pool.schedule(boost::bind(&ParallelAggregatingBlockInputStream::calculate, this, boost::ref(children[i]), boost::ref(*many_data[i]), boost::ref(exceptions[i])));
+			pool.schedule(boost::bind(&ParallelAggregatingBlockInputStream::calculate, this,
+				boost::ref(children[i]), boost::ref(*many_data[i]), boost::ref(exceptions[i]), current_memory_tracker));
 		}
 		pool.wait();
 
@@ -97,8 +98,10 @@ private:
 	boost::threadpool::pool pool;
 
 	/// Вычисления, которые выполняются в отдельном потоке
-	void calculate(BlockInputStreamPtr & input, AggregatedDataVariants & data, ExceptionPtr & exception)
+	void calculate(BlockInputStreamPtr & input, AggregatedDataVariants & data, ExceptionPtr & exception, MemoryTracker * memory_tracker)
 	{
+		current_memory_tracker = memory_tracker;
+
 		try
 		{
 			aggregator->execute(input, data);
diff --git a/dbms/include/DB/DataStreams/UnionBlockInputStream.h b/dbms/include/DB/DataStreams/UnionBlockInputStream.h
index dcd57371aa7..35acb5b0a35 100644
--- a/dbms/include/DB/DataStreams/UnionBlockInputStream.h
+++ b/dbms/include/DB/DataStreams/UnionBlockInputStream.h
@@ -148,7 +148,7 @@ protected:
 			threads_data.resize(max_threads);
 			for (ThreadsData::iterator it = threads_data.begin(); it != threads_data.end(); ++it)
 			{
-				it->runnable = new Thread(*this);
+				it->runnable = new Thread(*this, current_memory_tracker);
 				it->thread = new Poco::Thread;
 				it->thread->start(*it->runnable);
 			}
@@ -202,7 +202,10 @@ private:
 	class Thread : public Poco::Runnable
 	{
 	public:
-		Thread(UnionBlockInputStream & parent_) : parent(parent_) {}
+		Thread(UnionBlockInputStream & parent_, MemoryTracker * memory_tracker) : parent(parent_)
+		{
+			current_memory_tracker = memory_tracker;
+		}
 
 		void run()
 		{
diff --git a/dbms/include/DB/IO/BufferWithOwnMemory.h b/dbms/include/DB/IO/BufferWithOwnMemory.h
index 3791463c4bb..144acdfd853 100644
--- a/dbms/include/DB/IO/BufferWithOwnMemory.h
+++ b/dbms/include/DB/IO/BufferWithOwnMemory.h
@@ -3,6 +3,7 @@
 #include <boost/noncopyable.hpp>
 
 #include <DB/Common/ProfileEvents.h>
+#include <DB/Common/MemoryTracker.h>
 
 #include <DB/Core/Exception.h>
 #include <DB/Core/ErrorCodes.h>
@@ -87,6 +88,9 @@ private:
 		ProfileEvents::increment(ProfileEvents::IOBufferAllocs);
 		ProfileEvents::increment(ProfileEvents::IOBufferAllocBytes, m_capacity);
 
+		if (current_memory_tracker)
+			current_memory_tracker->alloc(m_capacity);
+
 		if (!alignment)
 		{
 			m_data = reinterpret_cast<char *>(malloc(m_capacity));
@@ -105,8 +109,13 @@ private:
 
 	void dealloc()
 	{
-		if (m_data)
-			free(reinterpret_cast<void *>(m_data));
+		if (!m_data)
+			return;
+
+		free(reinterpret_cast<void *>(m_data));
+
+		if (current_memory_tracker)
+			current_memory_tracker->free(m_capacity);
 	}
 };
 
diff --git a/dbms/include/DB/Interpreters/Limits.h b/dbms/include/DB/Interpreters/Limits.h
index 03b71be6eb2..1fe7d7f6424 100644
--- a/dbms/include/DB/Interpreters/Limits.h
+++ b/dbms/include/DB/Interpreters/Limits.h
@@ -72,7 +72,10 @@ struct Limits
 	/** Ограничения для максимального размера запоминаемого состояния при выполнении DISTINCT. */ \
 	M(SettingUInt64, max_rows_in_distinct, 0) \
 	M(SettingUInt64, max_bytes_in_distinct, 0) \
-	M(SettingOverflowMode<false>, distinct_overflow_mode, OverflowMode::THROW)
+	M(SettingOverflowMode<false>, distinct_overflow_mode, OverflowMode::THROW) \
+	\
+	/** Максимальное использование памяти при обработке запроса. 0 - не ограничено. */ \
+	M(SettingUInt64, max_memory_usage, 0) \
 
 #define DECLARE(TYPE, NAME, DEFAULT) \
 	TYPE NAME {DEFAULT};
diff --git a/dbms/include/DB/Interpreters/ProcessList.h b/dbms/include/DB/Interpreters/ProcessList.h
index 8792ab1eab8..09db7c14a2a 100644
--- a/dbms/include/DB/Interpreters/ProcessList.h
+++ b/dbms/include/DB/Interpreters/ProcessList.h
@@ -10,6 +10,7 @@
 #include <DB/Core/Defines.h>
 #include <DB/Core/Exception.h>
 #include <DB/Core/ErrorCodes.h>
+#include <DB/Common/MemoryTracker.h>
 #include <DB/IO/WriteHelpers.h>
 
 
@@ -34,15 +35,26 @@ public:
 		
 		Stopwatch watch;
 
-		volatile size_t rows_processed;
-		volatile size_t bytes_processed;
+		volatile size_t rows_processed = 0;
+		volatile size_t bytes_processed = 0;
 
-		bool is_cancelled;
+		MemoryTracker memory_tracker;
+
+		bool is_cancelled = false;
 		
 
-		Element(const String & query_, const String & user_, const String & query_id_, const Poco::Net::IPAddress & ip_address_)
-			: query(query_), user(user_), query_id(query_id_), ip_address(ip_address_), rows_processed(0), bytes_processed(0),
-			is_cancelled(false) {}
+		Element(const String & query_, const String & user_,
+			const String & query_id_, const Poco::Net::IPAddress & ip_address_,
+			size_t max_memory_usage)
+			: query(query_), user(user_), query_id(query_id_), ip_address(ip_address_), memory_tracker(max_memory_usage)
+		{
+			current_memory_tracker = &memory_tracker;
+		}
+
+		~Element()
+		{
+			current_memory_tracker = nullptr;
+		}
 
 		bool update(size_t rows, size_t bytes) volatile
 		{
@@ -113,7 +125,7 @@ public:
 	  * Если времени не хватило - кинуть исключение.
 	  */
 	EntryPtr insert(const String & query_, const String & user_, const String & query_id_, const Poco::Net::IPAddress & ip_address_,
-		size_t max_wait_milliseconds = DEFAULT_QUERIES_QUEUE_WAIT_TIME_MS, bool replace_running_query = false)
+		size_t max_memory_usage = 0, size_t max_wait_milliseconds = DEFAULT_QUERIES_QUEUE_WAIT_TIME_MS, bool replace_running_query = false)
 	{
 		EntryPtr res;
 
@@ -134,7 +146,7 @@ public:
 					{
 						if (!replace_running_query)
 							throw Exception("Query with id = " + query_id_ + " is already running.",
-											ErrorCodes::QUERY_WITH_SAME_ID_IS_ALREADY_RUNNING);
+								ErrorCodes::QUERY_WITH_SAME_ID_IS_ALREADY_RUNNING);
 						element->second->is_cancelled = true;
 						/// В случае если запрос отменяется, данные о нем удаляются из мапа в момент отмены.
 						queries->second.erase(element);
@@ -144,7 +156,7 @@ public:
 
 			++cur_size;
 
-			res = new Entry(*this, cont.insert(cont.end(), Element(query_, user_, query_id_, ip_address_)));
+			res = new Entry(*this, cont.emplace(cont.end(), query_, user_, query_id_, ip_address_, max_memory_usage));
 
 			if (!query_id_.empty())
 				user_to_queries[user_][query_id_] = &res->get();
diff --git a/dbms/include/DB/Interpreters/SplittingAggregator.h b/dbms/include/DB/Interpreters/SplittingAggregator.h
index b4d7fd118d9..c52c78509e8 100644
--- a/dbms/include/DB/Interpreters/SplittingAggregator.h
+++ b/dbms/include/DB/Interpreters/SplittingAggregator.h
@@ -91,9 +91,9 @@ private:
 	/// Для более точного контроля max_rows_to_group_by.
 	size_t size_of_all_results;
 
-	void calculateHashesThread(Block & block, size_t begin, size_t end, ExceptionPtr & exception);
-	void aggregateThread(Block & block, AggregatedDataVariants & result, size_t thread_no, ExceptionPtr & exception);
-	void convertToBlockThread(AggregatedDataVariants & data_variant, Block & block, bool final, ExceptionPtr & exception);
+	void calculateHashesThread(Block & block, size_t begin, size_t end, ExceptionPtr & exception, MemoryTracker * memory_tracker);
+	void aggregateThread(Block & block, AggregatedDataVariants & result, size_t thread_no, ExceptionPtr & exception, MemoryTracker * memory_tracker);
+	void convertToBlockThread(AggregatedDataVariants & data_variant, Block & block, bool final, ExceptionPtr & exception, MemoryTracker * memory_tracker);
 };
 
 
diff --git a/dbms/include/DB/Interpreters/executeQuery.h b/dbms/include/DB/Interpreters/executeQuery.h
index cd7c532935a..e776f8dfed9 100644
--- a/dbms/include/DB/Interpreters/executeQuery.h
+++ b/dbms/include/DB/Interpreters/executeQuery.h
@@ -12,11 +12,11 @@ namespace DB
 /** Парсит и исполняет запрос.
   */
 void executeQuery(
-	ReadBuffer & istr,								/// Откуда читать запрос (а также данные для INSERT-а, если есть)
-	WriteBuffer & ostr,								/// Куда писать результат
-	Context & context,								/// БД, таблицы, типы данных, движки таблиц, функции, агрегатные функции...
-	BlockInputStreamPtr & query_plan,				/// Сюда может быть записано описание, как выполнялся запрос
-	bool internal = false,							/// Если true - значит запрос порождён из другого запроса, и не нужно его регистировать в ProcessList-е.
+	ReadBuffer & istr,					/// Откуда читать запрос (а также данные для INSERT-а, если есть)
+	WriteBuffer & ostr,					/// Куда писать результат
+	Context & context,					/// БД, таблицы, типы данных, движки таблиц, функции, агрегатные функции...
+	BlockInputStreamPtr & query_plan,	/// Сюда может быть записано описание, как выполнялся запрос
+	bool internal = false,				/// Если true - значит запрос порождён из другого запроса, и не нужно его регистировать в ProcessList-е.
 	QueryProcessingStage::Enum stage = QueryProcessingStage::Complete);	/// До какой стадии выполнять SELECT запрос.
 
 
@@ -30,7 +30,7 @@ void executeQuery(
   * - затем читайте результат из BlockIO::in;
   *
   * Если запрос не предполагает записи данных или возврата результата, то out и in,
-  *  соответственно, будут равны NULL.
+  *  соответственно, будут равны nullptr.
   *
   * Часть запроса по парсингу и форматированию (секция FORMAT) необходимо выполнить отдельно.
   */
diff --git a/dbms/src/Common/MemoryTracker.cpp b/dbms/src/Common/MemoryTracker.cpp
new file mode 100644
index 00000000000..eb8097bcb88
--- /dev/null
+++ b/dbms/src/Common/MemoryTracker.cpp
@@ -0,0 +1,31 @@
+#include <Yandex/likely.h>
+#include <Yandex/logger_useful.h>
+#include <DB/Core/Exception.h>
+#include <DB/IO/WriteHelpers.h>
+
+#include <DB/Common/MemoryTracker.h>
+
+
+MemoryTracker::~MemoryTracker()
+{
+	LOG_DEBUG(&Logger::get("MemoryTracker"), "Peak memory usage for query: " << peak << " bytes.");
+}
+
+void MemoryTracker::alloc(Int32 size)
+{
+	Int32 will_be = __sync_add_and_fetch(&amount, size);
+
+	if (unlikely(limit && will_be > limit))
+	{
+		free(size);
+		throw DB::Exception("Memory limit exceeded: would use " + DB::toString(will_be) + " bytes"
+			" (attempt to allocate chunk of " + DB::toString(size) + " bytes)"
+			", maximum: " + DB::toString(limit) + " bytes", DB::ErrorCodes::MEMORY_LIMIT_EXCEEDED);
+	}
+
+	if (will_be > peak)
+		peak = will_be;
+}
+
+
+__thread MemoryTracker * current_memory_tracker = nullptr;
diff --git a/dbms/src/Interpreters/SplittingAggregator.cpp b/dbms/src/Interpreters/SplittingAggregator.cpp
index bc76d027010..bf80db339f0 100644
--- a/dbms/src/Interpreters/SplittingAggregator.cpp
+++ b/dbms/src/Interpreters/SplittingAggregator.cpp
@@ -102,7 +102,8 @@ void SplittingAggregator::execute(BlockInputStreamPtr stream, ManyAggregatedData
 				boost::ref(block),
 				rows * thread_no / threads,
 				rows * (thread_no + 1) / threads,
-				boost::ref(exceptions[thread_no])));
+				boost::ref(exceptions[thread_no]),
+				current_memory_tracker));
 
 		pool.wait();
 
@@ -120,7 +121,8 @@ void SplittingAggregator::execute(BlockInputStreamPtr stream, ManyAggregatedData
 				boost::ref(block),
 				boost::ref(*results[thread_no]),
 				thread_no,
-				boost::ref(exceptions[thread_no])));
+				boost::ref(exceptions[thread_no]),
+				current_memory_tracker));
 
 		pool.wait();
 
@@ -158,7 +160,8 @@ void SplittingAggregator::convertToBlocks(ManyAggregatedDataVariants & data_vari
 			boost::ref(*data_variants[thread_no]),
 			boost::ref(blocks[thread_no]),
 			final,
-			boost::ref(exceptions[thread_no])));
+			boost::ref(exceptions[thread_no]),
+			current_memory_tracker));
 
 	pool.wait();
 
@@ -166,8 +169,10 @@ void SplittingAggregator::convertToBlocks(ManyAggregatedDataVariants & data_vari
 }
 
 
-void SplittingAggregator::calculateHashesThread(Block & block, size_t begin, size_t end, ExceptionPtr & exception)
+void SplittingAggregator::calculateHashesThread(Block & block, size_t begin, size_t end, ExceptionPtr & exception, MemoryTracker * memory_tracker)
 {
+	current_memory_tracker = memory_tracker;
+
 	try
 	{
 		if (method == AggregatedDataVariants::KEY_64)
@@ -237,8 +242,11 @@ void SplittingAggregator::calculateHashesThread(Block & block, size_t begin, siz
 }
 
 
-void SplittingAggregator::aggregateThread(Block & block, AggregatedDataVariants & result, size_t thread_no, ExceptionPtr & exception)
+void SplittingAggregator::aggregateThread(
+	Block & block, AggregatedDataVariants & result, size_t thread_no, ExceptionPtr & exception, MemoryTracker * memory_tracker)
 {
+	current_memory_tracker = memory_tracker;
+
 	try
 	{
 		result.aggregator = this;
@@ -419,8 +427,11 @@ void SplittingAggregator::aggregateThread(Block & block, AggregatedDataVariants
 }
 
 
-void SplittingAggregator::convertToBlockThread(AggregatedDataVariants & data_variant, Block & block, bool final, ExceptionPtr & exception)
+void SplittingAggregator::convertToBlockThread(
+	AggregatedDataVariants & data_variant, Block & block, bool final, ExceptionPtr & exception, MemoryTracker * memory_tracker)
 {
+	current_memory_tracker = memory_tracker;
+
 	try
 	{
 		block = convertToBlock(data_variant, final);
diff --git a/dbms/src/Interpreters/executeQuery.cpp b/dbms/src/Interpreters/executeQuery.cpp
index 320b12e269a..7b37baf8ef4 100644
--- a/dbms/src/Interpreters/executeQuery.cpp
+++ b/dbms/src/Interpreters/executeQuery.cpp
@@ -89,8 +89,10 @@ void executeQuery(
 	{
 		process_list_entry = context.getProcessList().insert(
 			query, context.getUser(), context.getCurrentQueryId(), context.getIPAddress(),
+			context.getSettingsRef().limits.max_memory_usage,
 			context.getSettingsRef().queue_max_wait_ms.totalMilliseconds(),
 			context.getSettingsRef().replace_running_query);
+
 		context.setProcessListElement(&process_list_entry->get());
 	}
 
@@ -159,8 +161,10 @@ BlockIO executeQuery(
 	{
 		process_list_entry = context.getProcessList().insert(
 			query, context.getUser(), context.getCurrentQueryId(), context.getIPAddress(),
+			context.getSettingsRef().limits.max_memory_usage,
 			context.getSettingsRef().queue_max_wait_ms.totalMilliseconds(),
 			context.getSettingsRef().replace_running_query);
+
 		context.setProcessListElement(&process_list_entry->get());
 	}
 
diff --git a/dbms/src/Storages/StorageSystemProcesses.cpp b/dbms/src/Storages/StorageSystemProcesses.cpp
index 5ddaaca0a03..da4ba7de8f7 100644
--- a/dbms/src/Storages/StorageSystemProcesses.cpp
+++ b/dbms/src/Storages/StorageSystemProcesses.cpp
@@ -12,13 +12,14 @@ namespace DB
 StorageSystemProcesses::StorageSystemProcesses(const std::string & name_, const Context & context_)
 	: name(name_), context(context_)
 {
-	columns.push_back(NameAndTypePair("user", 		new DataTypeString));
-	columns.push_back(NameAndTypePair("address",	new DataTypeString));
-	columns.push_back(NameAndTypePair("elapsed", 	new DataTypeFloat64));
-	columns.push_back(NameAndTypePair("rows_read", 	new DataTypeUInt64));
-	columns.push_back(NameAndTypePair("bytes_read",	new DataTypeUInt64));
-	columns.push_back(NameAndTypePair("query", 		new DataTypeString));
-	columns.push_back(NameAndTypePair("query_id", 	new DataTypeString));
+	columns.push_back(NameAndTypePair("user", 			new DataTypeString));
+	columns.push_back(NameAndTypePair("address",		new DataTypeString));
+	columns.push_back(NameAndTypePair("elapsed", 		new DataTypeFloat64));
+	columns.push_back(NameAndTypePair("rows_read", 		new DataTypeUInt64));
+	columns.push_back(NameAndTypePair("bytes_read",		new DataTypeUInt64));
+	columns.push_back(NameAndTypePair("memory_usage",	new DataTypeUInt64));
+	columns.push_back(NameAndTypePair("query", 			new DataTypeString));
+	columns.push_back(NameAndTypePair("query_id", 		new DataTypeString));
 }
 
 StoragePtr StorageSystemProcesses::create(const std::string & name_, const Context & context_)
@@ -66,6 +67,12 @@ BlockInputStreams StorageSystemProcesses::read(
 	col_bytes_read.column = new ColumnUInt64;
 	block.insert(col_bytes_read);
 
+	ColumnWithNameAndType col_memory_usage;
+	col_memory_usage.name = "memory_usage";
+	col_memory_usage.type = new DataTypeUInt64;
+	col_memory_usage.column = new ColumnUInt64;
+	block.insert(col_memory_usage);
+
 	ColumnWithNameAndType col_query;
 	col_query.name = "query";
 	col_query.type = new DataTypeString;
@@ -90,6 +97,7 @@ BlockInputStreams StorageSystemProcesses::read(
 		col_elapsed.column->insert(it->watch.elapsedSeconds());
 		col_rows_read.column->insert(rows_read);
 		col_bytes_read.column->insert(bytes_read);
+		col_memory_usage.column->insert(UInt64(it->memory_tracker.get()));
 		col_query.column->insert(it->query);
 		col_query_id.column->insert(it->query_id);
 	}