► (3) 新数据更有价值 ► (4) 数据总是随时间变化而不断变化 Why we choose it ► 解决方案 ► (1) Row-Orient Database ► (2) Column-Orient Database ► (3) Time-Series-Orient Database Why we choose it Time Name Age Humidity HeartRate WHERE Time BETWEEN ... AND ... AND Name = “Tom” Red : Data needed Green : Data Scaned ► Column-Orient Database Why we choose it Temperature 11 20 ... 11 21 Time 2019/10/10/ 10:00:00 “好的”我们都想要 ! Why we choose it How we do ► ClickHouse 实现方式 ► (1) Column-Orient Model ► (2) Time-Series-Orient Model How we do ► Column-Orient Model How we do CREATE TABLE demonstration.insert_view

基于位图的分布式计算引擎 API Server Scheduler SQL-Parser QueryOptimier Column1 DataNode Column2 Column3 ColumnN Column1 DataNode Column2 Column3 ColumnN bitmap 画像下钻分布式计算引擎 多维 提取 iData大数据分析引擎 分布式多维计算引擎

器上面最大内存使用量 2：除了些简单的SQL，空间复杂度是O(1) 如: select count(1) from table where column=value select column1, column2 from table where column=value 凡是涉及group by, order by, distinct, join这样的SQL内存占用不再是O(1) 解决： 1：

Extended(X) • We prefer the largest native numerics, i.e. UInt64, Int64, or Float64 • Data types vs. column types • Furthermore? • The real arraySum also supports lambda Begin Content Area = 16,30 24 24 Scalar Function Implementation Then... arraySum([1, 2, 3]) • Data are handled per block • Column type of the argument = ColumnArray(ColumnVector(X), ColumnOffset) • ColumnConst can be handled explicitly Content Area = 16,30 28 Zora: High-performance Algorithm Implementation Framework Main Concepts Column-oriented & Memory Densed High memory efficency and avoiding unnecessary IO Smooth integration with

streamToClickHouse(clickHouseServers, 8123, "default", “password", “testDB", Seq(new Column("metric"), new Column("domain")) ��������������� ���������� ���������� ������������������ ������������������

GROUP BY id HAVING sum(sign)>0 ○ Need to use GROUP BY in every query ○ Not suitable for multi-column primary key Our Solution: PTS Key Features ● Only one config file needed for a new Clickhouse Bob1 2 2 Bob2 1 2 Bob2 3 id name flag 2 Bob1 2 2 Bob2 2 2 Bob2 3 PTS Magical Flag Add new column PTS Magical Flag Why fast? Clickhouse Mutation PTS Magical Flag Why fast? Clickhouse Mutation

Нельзя писать в сервер, отрезанный от кворума ZK Репликация с точки зрения CAP–теоремы Всё вместе › Column–oriented › Сверхбыстрые интерактивные запросы › Диалект SQL + расширения › Плохо подходит для

sumState(number) FROM numbers(5) FORMAT TSV \n\0\0\0\0\0\0\0 -- Binary data. ASCII for \n is 10. Column has special type SELECT toTypeName(sumState(number)) FROM numbers(5) FORMAT TSV AggregateFunction(sum

sumState(number) FROM numbers(5) FORMAT TSV \n\0\0\0\0\0\0\0 -- Binary data. ASCII for \n is 10. Column has special type SELECT toTypeName(sumState(number)) FROM numbers(5) FORMAT TSV AggregateFunction(sum

contrib static clang-7 release address contrib static clang-7 release undefined contrib static clang-7 release — contrib shared clang-8 release thread contrib static gcc-8 release — contrib static gcc-8 gcc-8 release — system static И это не все... 11 / 77 Тестирование ClickHouse, которого мы заслуживаем ClickHouse не тормозит, сборка тормозит › Время одной сборки – 20 минут на 32 ядрах + 128 RAM ›