better code in ReaderCompact and perf test added

src/Storages/MergeTree/MergeTreeReaderCompact.cpp

@@ -64,7 +64,9 @@ MergeTreeReaderCompact::MergeTreeReaderCompact(
         column_positions[i] = std::move(position);
     }
 
-    auto buffer_size = getReadBufferSize();
+    /// Do not use max_read_buffer_size, but try to lower buffer size with maximal size of granule to avoid reading much data.
+    auto buffer_size = getReadBufferSize(data_part, marks_loader, column_positions, all_mark_ranges);
+    std::cerr << "buffer_size: " << buffer_size << "\n";
     if (!buffer_size || settings.max_read_buffer_size < buffer_size)
         buffer_size = settings.max_read_buffer_size;
 
@@ -240,47 +242,90 @@ bool MergeTreeReaderCompact::isContinuousReading(size_t mark, size_t column_posi
         || (mark == last_mark + 1 && column_position == 0 && last_column == data_part->getColumns().size() - 1);
 }
 
-
+namespace
-size_t MergeTreeReaderCompact::getReadBufferSize()
 {
-    size_t buffer_size = 0;
-    size_t columns_num = columns.size();
-    size_t last_column_position = data_part->getColumns().size() - 1;
-    size_t file_size = data_part->getFileSizeOrZero(MergeTreeDataPartCompact::DATA_FILE_NAME_WITH_EXTENSION);
-    size_t marks_count = data_part->getMarksCount();
 
-    auto next = [&](size_t & row_index, size_t & column_index)
+/// A simple class that helps to iterate over 2-dim marks of compact parts.
+class MarksCounter
+{
+public:
+    MarksCounter(size_t rows_num_, size_t columns_num_)
+        : rows_num(rows_num_), columns_num(columns_num_) {}
+
+    struct Iterator
     {
-        if (column_index == last_column_position)
+        size_t row;
-        {
+        size_t column;
-            ++row_index;
+        MarksCounter * counter;
-            column_index = 0;
-            if (row_index == marks_count)
-                return false;
-        }
-        else
-            ++column_index;
 
-        return true;
+        Iterator(size_t row_, size_t column_, MarksCounter * counter_)
+            : row(row_), column(column_), counter(counter_) {}
+
+        Iterator operator++()
+        {
+            if (column + 1 == counter->columns_num)
+            {
+                ++row;
+                column = 0;
+            }
+            else
+            {
+                ++column;
+            }
+
+            return *this;
+        }
+
+        Iterator operator++(int)
+        {
+            auto tmp = *this;
+            ++*this;
+            return tmp;
+        }
+
+        bool operator==(const Iterator & other) const { return row == other.row && column == other.column; }
+        bool operator!=(const Iterator & other) const { return !(*this == other); }
     };
 
-    for (const auto & mark_range : all_mark_ranges)
+    Iterator get(size_t row, size_t column) { return Iterator(row, column, this); }
+    Iterator begin() { return get(0, 0); }
+    Iterator end() { return get(rows_num, 0); }
+
+private:
+    size_t rows_num;
+    size_t columns_num;
+};
+
+}
+
+size_t MergeTreeReaderCompact::getReadBufferSize(
+    const DataPartPtr & data_part,
+    MergeTreeMarksLoader & marks_loader,
+    const ColumnPositions & column_positions,
+    const MarkRanges & mark_ranges)
+{
+    size_t buffer_size = 0;
+    size_t columns_num = column_positions.size();
+    size_t file_size = data_part->getFileSizeOrZero(MergeTreeDataPartCompact::DATA_FILE_NAME_WITH_EXTENSION);
+
+    MarksCounter counter(data_part->getMarksCount(), data_part->getColumns().size());
+
+    for (const auto & mark_range : mark_ranges)
     {
-        for (size_t mark = mark_range.begin; mark <= mark_range.end; ++mark)
+        for (size_t mark = mark_range.begin; mark < mark_range.end; ++mark)
         {
             for (size_t i = 0; i < columns_num; ++i)
             {
                 if (!column_positions[i])
                     continue;
 
-                size_t row_ind = mark;
+                auto it = counter.get(mark, *column_positions[i]);
-                size_t col_ind = *column_positions[i];
+                size_t cur_offset = marks_loader.getMark(it.row, it.column).offset_in_compressed_file;
-                size_t cur_offset = marks_loader.getMark(mark, col_ind).offset_in_compressed_file;
 
-                while (next(row_ind, col_ind) && cur_offset == marks_loader.getMark(row_ind, col_ind).offset_in_compressed_file)
+                while (it != counter.end() && cur_offset == marks_loader.getMark(it.row, it.column).offset_in_compressed_file)
-                    ;
+                    ++it;
 
-                size_t next_offset = (row_ind == marks_count ? file_size : marks_loader.getMark(row_ind, col_ind).offset_in_compressed_file);
+                size_t next_offset = (it == counter.end() ? file_size : marks_loader.getMark(it.row, it.column).offset_in_compressed_file);
                 buffer_size = std::max(buffer_size, next_offset - cur_offset);
             }
         }

src/Storages/MergeTree/MergeTreeReaderCompact.h

@@ -10,6 +10,9 @@ namespace DB
 class MergeTreeDataPartCompact;
 using DataPartCompactPtr = std::shared_ptr<const MergeTreeDataPartCompact>;
 
+class IMergeTreeDataPart;
+using DataPartPtr = std::shared_ptr<const IMergeTreeDataPart>;
+
 /// Reader for compact parts
 class MergeTreeReaderCompact : public IMergeTreeReader
 {
@@ -42,7 +45,8 @@ private:
     MergeTreeMarksLoader marks_loader;
 
     /// Positions of columns in part structure.
-    std::vector<ColumnPosition> column_positions;
+    using ColumnPositions = std::vector<ColumnPosition>;
+    ColumnPositions column_positions;
     /// Should we read full column or only it's offsets
     std::vector<bool> read_only_offsets;
 
@@ -54,7 +58,13 @@ private:
     void readData(const String & name, IColumn & column, const IDataType & type,
         size_t from_mark, size_t column_position, size_t rows_to_read, bool only_offsets = false);
 
-    size_t getReadBufferSize();
+    /// Returns maximal value of granule size in compressed file from @mark_ranges.
+    /// This value is used as size of read buffer.
+    static size_t getReadBufferSize(
+        const DataPartPtr & data_part,
+        MergeTreeMarksLoader & marks_loader,
+        const ColumnPositions & column_positions,
+        const MarkRanges & mark_ranges);
 };
 
 }

tests/performance/read_from_comp_parts.xml

@@ -0,0 +1,23 @@
+<test>
+    <create_query>
+        CREATE TABLE mt_comp_parts
+        ENGINE = MergeTree
+        ORDER BY (c1, c2)
+        SETTINGS min_rows_for_wide_part = 1000000000 AS
+        SELECT *
+        FROM generateRandom('c1 UInt32, c2 UInt64, s1 String, arr1 Array(UInt32), c3 UInt64, s2 String', 0, 30, 30)
+        LIMIT 50000000
+    </create_query>
+
+    <settings>
+        <max_threads>8</max_threads>
+    </settings>
+
+    <query>SELECT count() FROM mt_comp_parts WHERE NOT ignore(c1)</query>
+    <query>SELECT count() FROM mt_comp_parts WHERE NOT ignore(c2, s1, arr1, s2)</query>
+    <query>SELECT count() FROM mt_comp_parts WHERE NOT ignore(c1, s1, c3)</query>
+    <query>SELECT count() FROM mt_comp_parts WHERE NOT ignore(c1, c2, c3)</query>
+    <query>SELECT count() FROM mt_comp_parts WHERE NOT ignore(*)</query>
+
+    <drop_query>DROP TABLE IF EXISTS mt_comp_parts</drop_query>
+</test>