{"type":"doc","content":[{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"ClickHouse"},{"type":"text","text":" 是俄罗斯最大的搜索引擎Yandex在2016年开源的数据库管理系统（DBMS），主要用于联机分析处理（OLAP）。其采用了面向列的存储方式，性能远超传统面向行的DBMS，近几年受到广泛关注。"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"本文将介绍 "},{"type":"text","marks":[{"type":"strong"}],"text":"ClickHouse MergeTree系列表引擎"},{"type":"text","text":" 的相关知识，并通过示例分析MergeTree存储引擎的数据存储结构。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"1 MergeTree表引擎简介"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"MergeTree（合并树）系列表引擎是ClickHouse提供的最具特色的存储引擎。MergeTree引擎支持数据按主键、数据分区、数据副本以及数据采样等特性。官方提供了包括MergeTree、ReplacingMergeTree、SummingMergeTree、AggregatingMergeTree、CollapsingMergeTree、VersionedCollapsingMergeTree、GraphiteMergeTree等7种不同类型的MergeTree引擎的实现，以及与其相对应的支持数据副本的MergeTree引擎（Replicated*）。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/e6\/15\/e6940e6b994e7df1fc72be8328d18a15.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"首先来介绍一下 "},{"type":"text","marks":[{"type":"strong"}],"text":"MergeTree核心引擎"},{"type":"text","text":"："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"ReplacingMergeTree"},{"type":"text","text":"： 在后台数据合并期间，对具有相同排序键的数据进行去重操作。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"SummingMergeTree"},{"type":"text","text":"： 当合并数据时，会把具有相同主键的记录合并为一条记录。根据聚合字段设置，该字段的值为聚合后的汇总值，非聚合字段使用第一条记录的值，聚合字段类型必须为数值类型。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"AggregatingMergeTree"},{"type":"text","text":"： 在同一数据分区下，可以将具有相同主键的数据进行聚合。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"CollapsingMergeTree"},{"type":"text","text":"： 在同一数据分区下，对具有相同主键的数据进行折叠合并。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"VersionedCollapsingMergeTree"},{"type":"text","text":"："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"基于CollapsingMergeTree引擎，增添了数据版本信息字段配置选项。在数据依据ORDER BY设置对数据进行排序的基础上，如果数据的版本信息列不在排序字段中，那么版本信息会被隐式的作为ORDER BY的最后一列从而影响数据排序。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"GraphiteMergeTree"},{"type":"text","text":"： 用来存储时序数据库Graphites的数据。"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"MergeTree是该系列引擎中最核心的引擎，其他引擎均以MergeTree为基础，并在数据合并过程中实现了不同的特性，从而构成了MergeTree表引擎家族。下面我们通过MergeTree来具体了解MergeTree表系列引擎。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"2 MergeTree引擎"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"2.1 表创建"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"创建MergeTree的DDL如下所示："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"text"},"content":[{"type":"text","text":"CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster] \n(\n name1 [type1] [DEFAULT|MATERIALIZED|ALIAS expr1] [TTL expr1], \n name2 [type2] [DEFAULT|MATERIALIZED|ALIAS expr2] [TTL expr2], \n ... \n) ENGINE = MergeTree()\n ORDER BY expr \n [PARTITION BY expr] \n [PRIMARY KEY expr] \n [SAMPLE BY expr] \n [TTL expr [DELETE|TO DISK 'xxx'|TO VOLUME 'xxx'], ...] \n [SETTINGS name=value, ...\n"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"这里说明一下MergeTree引擎的主要参数："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"必填选项"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"ENGINE"},{"type":"text","text":" ：引擎名字，MergeTree引擎无参数。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"ORDER BY"},{"type":"text","text":" ：排序键，可以由一列或多列组成，决定了数据以何种方式进行排序，例如ORDER BY（CounterID, EventDate）。如果没有显示指定PRIMARY KEY，那么将使用ORDER BY作为PRIMARY KEY。通常只指定ORDER BY即可。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"选填选项"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"PARTITION BY"},{"type":"text","text":" ：分区键，指明表中的数据以何种规则进行分区。分区是在一个表中通过指定的规则划分而成的逻辑数据集。分区可以按任意标准进行，如按月、按日或按事件类型。为了减少需要操作的数据，每个分区都是分开存储的。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"PRIMARY KEY"},{"type":"text","text":" ：主键，设置后会按照主键生成一级索引（primary.idx），数据会依据索引的设置进行排序，从而加速查询性能。默认情况下，PRIMARY KEY与ORDER BY设置相同，所以通常情况下直接使用ORDER BY设置来替代主键设置。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"SAMPLE BY"},{"type":"text","text":" ：数据采样设置，如果显示配置了该选项，那么主键配置中也应该包括此配置。例如 ORDER BY CounterID \/ EventDate \/ intHash32（UserID）、SAMPLE BY intHash32（UserID）。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"TTL"},{"type":"text","text":" ：数据存活时间，可以为某一字段列或者一整张表设置TTL，设置中必须包含Date或DateTime字段类型。如果设置在列上，那么会删除字段中过期的数据。如果设置的是表级的TTL，那么会删除表中过期的数据。如果设置了两种类型，那么按先到期的为准。例如，TTL createtime + INTERVAL 1 DAY，即一天后过期。使用场景包括定期删除数据，或者定期将数据进行归档。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"index_granularity"},{"type":"text","text":" ：索引间隔粒度。MergeTree索引为稀疏索引，每index_granularity个数据产生一条索引。index_granularity默认设置为8092。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"enable_mixed_granularity_parts"},{"type":"text","text":" ：是否启动index_granularity_bytes来控制索引粒度大小。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"index_granularity_bytes"},{"type":"text","text":" ：索引粒度，以字节为单位，默认10Mb。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"merge_max_block_size"},{"type":"text","text":" ：数据块合并最大记录个数，默认8192。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"merge_with_ttl_timeout"},{"type":"text","text":" ：合并频率最小时间间隔，默认1天。"}]}]}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"2.2 数据存储结构"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"首先创建一个test表，DDL如下:"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"text"},"content":[{"type":"text","text":"CREATE TABLE test.test \n( \n id UInt64, \n type UInt8, \n create_time DateTime \n) ENGINE = MergeTree() \n PARTITION BY toYYYYMMDD(create_time) \n ORDER BY (id) \n SETTINGS index_granularity = 4;\n"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"test表包括id、type、create等三个字段，其中以create_time日期字段作为分区键，并将日期格式转化为YYYYMMDD。按照id字段进行排序。由于没有显式设置主键，所以引擎默认使用ORDER BY设置的id列作为索引字段，并生成索引文件。index_granularity设置为4，意味着每4条数据产生一条索引数据。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"插入一条测试数据："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"text"},"content":[{"type":"text","text":"insert into test.test(id, type, create_time) VALUES (1, 1, toDateTime('2021-03-01 00:00:00'));\n"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"使用如下命令查看test表分区相关信息："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"text"},"content":[{"type":"text","text":" SELECT \n database, \n table, \n partition, \n partition_id, \n name, \n active, \n path \n FROM system.parts \n WHERE table = 'test' \n"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"返回结果如下图所示："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/43\/f1\/43f14146a075b8a09417b34ec1abyyf1.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"从上图中可以看到test表中返回了一条partitionid为20210301的数据分区的记录，从name字段中我们可以得知，此分区的目录名为20210301_8_8_0。20210301_8_8_0这个目录名字到底有什么含义呢？下面来介绍一下分区规则以及分区目录的命名规则。"}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"2.2.1 数据分区ID生成规则"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"数据分区规则由分区ID决定，分区ID由PARTITION BY分区键决定。根据分区键字段类型，ID生成规则可分为："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"未定义分区键"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"没有定义PARTITION BY，默认生成一个目录名为all的数据分区，所有数据均存放在all目录下。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"整型分区键"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"分区键为整型，那么直接用该整型值的字符串形式做为分区ID。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"日期类分区键"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"分区键为日期类型，或者可以转化成日期类型。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"其他类型分区键"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"String、Float类型等，通过128位的Hash算法取其Hash值作为分区ID。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"上面我们插入一条日期为2021-03-01 00:00:00的数据，对该字段格式化后生成的数据分区id就是20210301。"}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"2.2.2 数据分区目录命名规则"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"目录命名规则如下："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"text"},"content":[{"type":"text","text":"PartitionId_MinBlockNum_MaxBlockNum_Level\n"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"PartitionID"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"分区id，例如20210301。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"MinBlockNum"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"最小分区块编号，自增类型，从1开始向上递增。每产生一个新的目录分区就向上递增一个数字。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"MaxBlockNum"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"最大分区块编号，新创建的分区MinBlockNum等于MaxBlockNum的编号。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"Level"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"合并的层级，被合并的次数。合并次数越多，层级值越大。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/da\/b5\/da1e10c251c73085ddffe598db5e07b5.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"从上图可知，此分区的分区id为20210301，当前分区的MinBlockNum和MinBlockNum均为8，而level为0，表示此分区没有合并过。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"2.3 数据分区文件组织结构"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在了解了分区目录名字的生成规则后，下面来看看数据分区目录下的文件组织结构。以2021030188_0分区为例："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/6c\/f3\/6cc0b607ce0a9c35bc2aecc0e2e495f3.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"从图中可以看到，目录中的文件主要包括bin文件、mrk文件、primary.idx文件以及其他相关文件。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"bin文件"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"数据文件，存储的是某一列的数据。数据表中的每一列都对应一个与其字段名相同的bin文件，例如id.bin存储的是表test中id列的数据。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"mrk文件"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"标记文件，每一列都对应一个与其字段名相同的标记文件，标记文件在idx索引文件和bin数据文件之间起到了桥梁作用。以mrk2结尾的文件，表示该表启用了自适应索引间隔。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"primary.idx文件"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"主键索引文件，用于加快查询效率。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"count.txt"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"数据分区中数据总记录数。上述20210301_8_8_0的数据分区中，该文件中的记录总数为1。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"columns.txt"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"表中所有列数的信息，包括字段名和字段类型。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"partion.dat"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"用于保存分区表达式的值。上述20210301_8_8_0的数据分区中该文件中的值为20210301。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"minmax_create_time.idx"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"分区键的最大最小值。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"checksums.txt"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"校验文件，用于校验各个文件的正确性。存放各个文件的size以及hash值。"}]}]}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"2.3.1 数据文件"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"MergeTree中，每列都对应一个bin文件单独存放该列数据。例如，id.bin存放的是id列的数据。所有数据都经过数据压缩、排序，最后以数据块的形式写入bin文件中。bin中数据以压缩数据块为单位写入文件中。每个数据块由头信息和压缩数据组成。头部信息包括校验和、数据压缩算法、数据压缩前大小和压缩后大小组成。压缩数据由granule组成，granule大小与index_granularity相关。"}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"2.3.2 索引文件"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"MergeTree索引为稀疏索引，它并不索引单条数据，而是索引一定范围的数据。也就是从已排序的全量数据中，间隔性的选取一些数据记录主键字段的值来生成primary.idx索引文件，从而加快表查询效率。间隔设置参数为index_granularity。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/bb\/4c\/bbc302f044278bbfcc3c16d63f15a34c.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我们向表test中插入9条数据，"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"text"},"content":[{"type":"text","text":"insert into test.test(id, type, create_time) VALUES (1, 1, toDateTime('2021-03-01 00:00:00')); \ninsert into test.test(id, type, create_time) VALUES (1, 2, toDateTime('2021-03-01 00:00:00')); \ninsert into test.test(id, type, create_time) VALUES (1, 3, toDateTime('2021-03-01 00:00:00')); \ninsert into test.test(id, type, create_time) VALUES (2, 1, toDateTime('2021-03-01 00:00:00')); \ninsert into test.test(id, type, create_time) VALUES (2, 1, toDateTime('2021-03-01 00:00:00')); \ninsert into test.test(id, type, create_time) VALUES (3, 1, toDateTime('2021-03-01 00:00:00')); \ninsert into test.test(id, type, create_time) VALUES (3, 1, toDateTime('2021-03-01 00:00:00')); \ninsert into test.test(id, type, create_time) VALUES (4, 1, toDateTime('2021-03-01 00:00:00')); \ninsert into test.test(id, type, create_time) VALUES (5, 1, toDateTime('2021-03-01 00:00:00'));\n"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"因为index_granularity设置为4，所以每4条数据就会生成一条索引记录，即使用插入的第1、5、9条数据id字段的值生成索引文件记录。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/d6\/9c\/d6ef6c6d12d669a97b0940e852caf29c.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"2.3.3 标记文件"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"mrk标记文件在primary.idx索引文件和bin数据文件之间起到了桥梁作用。primary.idx文件中的每条索引在mrk文件中都有对应的一条记录。一条记录的组成包括："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"offset-compressed bin file"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"表示指向的压缩数据块在bin文件中的偏移量。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"offset-decompressed data block"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"表示指向的数据在解压数据块中的偏移量。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"row counts"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"代表数据记录行数，小于等于index_granularity所设置的值。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/57\/d6\/57c2c50d02c4bf7637007fb3b337d6d6.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"索引，标记和数据文件下图所示："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/b2\/85\/b2dabc9da1d48abca91627a46c08dd85.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"作者：TalkingData 张凯"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"参考文档："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"1.https:\/\/clickhouse.tech\/docs"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"2.http:\/\/www.clickhouse.com.cn\/topic\/5ffec51eba8f16b55dd0ffe4"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"3.《ClickHouse原理解析与应用实践》（机械工业出版社出版，作者朱凯）"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"本文转载自公众号TalkingData（ID：Talkingdata）。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"原文链接"},{"type":"text","text":"："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"link","attrs":{"href":"https:\/\/mp.weixin.qq.com\/s\/YWSmRqOOC3F5KeDBPg2g1A","title":"","type":null},"content":[{"type":"text","text":"ClickHouse核心引擎MergeTree解读"}]}]}]}
ClickHouse核心引擎MergeTree解读
{"type":"doc","content":[{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"ClickHouse"},{"type":"text","text":" 是俄罗斯最大的搜索引擎Yandex在2016年开源的数据库管理系统（DBMS），主要用于联机分析处理（OLAP）。其采用了面向列的存储方式，性能远超传统面向行的DBMS，近几年受到广泛关注。"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"本文将介绍 "},{"type":"text","marks":[{"type":"strong"}],"text":"ClickHouse MergeTree系列表引擎"},{"type":"text","text":" 的相关知识，并通过示例分析MergeTree存储引擎的数据存储结构。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"1 MergeTree表引擎简介"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"MergeTree（合并树）系列表引擎是ClickHouse提供的最具特色的存储引擎。MergeTree引擎支持数据按主键、数据分区、数据副本以及数据采样等特性。官方提供了包括MergeTree、ReplacingMergeTree、SummingMergeTree、AggregatingMergeTree、CollapsingMergeTree、VersionedCollapsingMergeTree、GraphiteMergeTree等7种不同类型的MergeTree引擎的实现，以及与其相对应的支持数据副本的MergeTree引擎（Replicated*）。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/e6\/15\/e6940e6b994e7df1fc72be8328d18a15.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"首先来介绍一下 "},{"type":"text","marks":[{"type":"strong"}],"text":"MergeTree核心引擎"},{"type":"text","text":"："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"ReplacingMergeTree"},{"type":"text","text":"： 在后台数据合并期间，对具有相同排序键的数据进行去重操作。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"SummingMergeTree"},{"type":"text","text":"： 当合并数据时，会把具有相同主键的记录合并为一条记录。根据聚合字段设置，该字段的值为聚合后的汇总值，非聚合字段使用第一条记录的值，聚合字段类型必须为数值类型。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"AggregatingMergeTree"},{"type":"text","text":"： 在同一数据分区下，可以将具有相同主键的数据进行聚合。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"CollapsingMergeTree"},{"type":"text","text":"： 在同一数据分区下，对具有相同主键的数据进行折叠合并。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"VersionedCollapsingMergeTree"},{"type":"text","text":"："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"基于CollapsingMergeTree引擎，增添了数据版本信息字段配置选项。在数据依据ORDER BY设置对数据进行排序的基础上，如果数据的版本信息列不在排序字段中，那么版本信息会被隐式的作为ORDER BY的最后一列从而影响数据排序。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"GraphiteMergeTree"},{"type":"text","text":"： 用来存储时序数据库Graphites的数据。"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"MergeTree是该系列引擎中最核心的引擎，其他引擎均以MergeTree为基础，并在数据合并过程中实现了不同的特性，从而构成了MergeTree表引擎家族。下面我们通过MergeTree来具体了解MergeTree表系列引擎。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"2 MergeTree引擎"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"2.1 表创建"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"创建MergeTree的DDL如下所示："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"text"},"content":[{"type":"text","text":"CREATE TABLE [IF NOT EXISTS] [db.]table_name [ON CLUSTER cluster] \n(\n name1 [type1] [DEFAULT|MATERIALIZED|ALIAS expr1] [TTL expr1], \n name2 [type2] [DEFAULT|MATERIALIZED|ALIAS expr2] [TTL expr2], \n ... \n) ENGINE = MergeTree()\n ORDER BY expr \n [PARTITION BY expr] \n [PRIMARY KEY expr] \n [SAMPLE BY expr] \n [TTL expr [DELETE|TO DISK 'xxx'|TO VOLUME 'xxx'], ...] \n [SETTINGS name=value, ...\n"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"这里说明一下MergeTree引擎的主要参数："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"必填选项"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"ENGINE"},{"type":"text","text":" ：引擎名字，MergeTree引擎无参数。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"ORDER BY"},{"type":"text","text":" ：排序键，可以由一列或多列组成，决定了数据以何种方式进行排序，例如ORDER BY（CounterID, EventDate）。如果没有显示指定PRIMARY KEY，那么将使用ORDER BY作为PRIMARY KEY。通常只指定ORDER BY即可。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"选填选项"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"PARTITION BY"},{"type":"text","text":" ：分区键，指明表中的数据以何种规则进行分区。分区是在一个表中通过指定的规则划分而成的逻辑数据集。分区可以按任意标准进行，如按月、按日或按事件类型。为了减少需要操作的数据，每个分区都是分开存储的。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"PRIMARY KEY"},{"type":"text","text":" ：主键，设置后会按照主键生成一级索引（primary.idx），数据会依据索引的设置进行排序，从而加速查询性能。默认情况下，PRIMARY KEY与ORDER BY设置相同，所以通常情况下直接使用ORDER BY设置来替代主键设置。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"SAMPLE BY"},{"type":"text","text":" ：数据采样设置，如果显示配置了该选项，那么主键配置中也应该包括此配置。例如 ORDER BY CounterID \/ EventDate \/ intHash32（UserID）、SAMPLE BY intHash32（UserID）。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"TTL"},{"type":"text","text":" ：数据存活时间，可以为某一字段列或者一整张表设置TTL，设置中必须包含Date或DateTime字段类型。如果设置在列上，那么会删除字段中过期的数据。如果设置的是表级的TTL，那么会删除表中过期的数据。如果设置了两种类型，那么按先到期的为准。例如，TTL createtime + INTERVAL 1 DAY，即一天后过期。使用场景包括定期删除数据，或者定期将数据进行归档。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"index_granularity"},{"type":"text","text":" ：索引间隔粒度。MergeTree索引为稀疏索引，每index_granularity个数据产生一条索引。index_granularity默认设置为8092。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"enable_mixed_granularity_parts"},{"type":"text","text":" ：是否启动index_granularity_bytes来控制索引粒度大小。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"index_granularity_bytes"},{"type":"text","text":" ：索引粒度，以字节为单位，默认10Mb。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"merge_max_block_size"},{"type":"text","text":" ：数据块合并最大记录个数，默认8192。"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"merge_with_ttl_timeout"},{"type":"text","text":" ：合并频率最小时间间隔，默认1天。"}]}]}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"2.2 数据存储结构"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"首先创建一个test表，DDL如下:"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"text"},"content":[{"type":"text","text":"CREATE TABLE test.test \n( \n id UInt64, \n type UInt8, \n create_time DateTime \n) ENGINE = MergeTree() \n PARTITION BY toYYYYMMDD(create_time) \n ORDER BY (id) \n SETTINGS index_granularity = 4;\n"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"test表包括id、type、create等三个字段，其中以create_time日期字段作为分区键，并将日期格式转化为YYYYMMDD。按照id字段进行排序。由于没有显式设置主键，所以引擎默认使用ORDER BY设置的id列作为索引字段，并生成索引文件。index_granularity设置为4，意味着每4条数据产生一条索引数据。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"插入一条测试数据："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"text"},"content":[{"type":"text","text":"insert into test.test(id, type, create_time) VALUES (1, 1, toDateTime('2021-03-01 00:00:00'));\n"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"使用如下命令查看test表分区相关信息："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"text"},"content":[{"type":"text","text":" SELECT \n database, \n table, \n partition, \n partition_id, \n name, \n active, \n path \n FROM system.parts \n WHERE table = 'test' \n"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"返回结果如下图所示："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/43\/f1\/43f14146a075b8a09417b34ec1abyyf1.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"从上图中可以看到test表中返回了一条partitionid为20210301的数据分区的记录，从name字段中我们可以得知，此分区的目录名为20210301_8_8_0。20210301_8_8_0这个目录名字到底有什么含义呢？下面来介绍一下分区规则以及分区目录的命名规则。"}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"2.2.1 数据分区ID生成规则"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"数据分区规则由分区ID决定，分区ID由PARTITION BY分区键决定。根据分区键字段类型，ID生成规则可分为："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"未定义分区键"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"没有定义PARTITION BY，默认生成一个目录名为all的数据分区，所有数据均存放在all目录下。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"整型分区键"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"分区键为整型，那么直接用该整型值的字符串形式做为分区ID。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"日期类分区键"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"分区键为日期类型，或者可以转化成日期类型。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"其他类型分区键"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"String、Float类型等，通过128位的Hash算法取其Hash值作为分区ID。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"上面我们插入一条日期为2021-03-01 00:00:00的数据，对该字段格式化后生成的数据分区id就是20210301。"}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"2.2.2 数据分区目录命名规则"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"目录命名规则如下："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"text"},"content":[{"type":"text","text":"PartitionId_MinBlockNum_MaxBlockNum_Level\n"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"PartitionID"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"分区id，例如20210301。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"MinBlockNum"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"最小分区块编号，自增类型，从1开始向上递增。每产生一个新的目录分区就向上递增一个数字。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"MaxBlockNum"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"最大分区块编号，新创建的分区MinBlockNum等于MaxBlockNum的编号。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"Level"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"合并的层级，被合并的次数。合并次数越多，层级值越大。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/da\/b5\/da1e10c251c73085ddffe598db5e07b5.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"从上图可知，此分区的分区id为20210301，当前分区的MinBlockNum和MinBlockNum均为8，而level为0，表示此分区没有合并过。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"2.3 数据分区文件组织结构"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在了解了分区目录名字的生成规则后，下面来看看数据分区目录下的文件组织结构。以2021030188_0分区为例："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/6c\/f3\/6cc0b607ce0a9c35bc2aecc0e2e495f3.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"从图中可以看到，目录中的文件主要包括bin文件、mrk文件、primary.idx文件以及其他相关文件。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"bin文件"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"数据文件，存储的是某一列的数据。数据表中的每一列都对应一个与其字段名相同的bin文件，例如id.bin存储的是表test中id列的数据。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"mrk文件"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"标记文件，每一列都对应一个与其字段名相同的标记文件，标记文件在idx索引文件和bin数据文件之间起到了桥梁作用。以mrk2结尾的文件，表示该表启用了自适应索引间隔。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"primary.idx文件"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"主键索引文件，用于加快查询效率。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"count.txt"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"数据分区中数据总记录数。上述20210301_8_8_0的数据分区中，该文件中的记录总数为1。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"columns.txt"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"表中所有列数的信息，包括字段名和字段类型。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"partion.dat"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"用于保存分区表达式的值。上述20210301_8_8_0的数据分区中该文件中的值为20210301。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"minmax_create_time.idx"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"分区键的最大最小值。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"checksums.txt"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"校验文件，用于校验各个文件的正确性。存放各个文件的size以及hash值。"}]}]}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"2.3.1 数据文件"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"MergeTree中，每列都对应一个bin文件单独存放该列数据。例如，id.bin存放的是id列的数据。所有数据都经过数据压缩、排序，最后以数据块的形式写入bin文件中。bin中数据以压缩数据块为单位写入文件中。每个数据块由头信息和压缩数据组成。头部信息包括校验和、数据压缩算法、数据压缩前大小和压缩后大小组成。压缩数据由granule组成，granule大小与index_granularity相关。"}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"2.3.2 索引文件"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"MergeTree索引为稀疏索引，它并不索引单条数据，而是索引一定范围的数据。也就是从已排序的全量数据中，间隔性的选取一些数据记录主键字段的值来生成primary.idx索引文件，从而加快表查询效率。间隔设置参数为index_granularity。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/bb\/4c\/bbc302f044278bbfcc3c16d63f15a34c.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我们向表test中插入9条数据，"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"text"},"content":[{"type":"text","text":"insert into test.test(id, type, create_time) VALUES (1, 1, toDateTime('2021-03-01 00:00:00')); \ninsert into test.test(id, type, create_time) VALUES (1, 2, toDateTime('2021-03-01 00:00:00')); \ninsert into test.test(id, type, create_time) VALUES (1, 3, toDateTime('2021-03-01 00:00:00')); \ninsert into test.test(id, type, create_time) VALUES (2, 1, toDateTime('2021-03-01 00:00:00')); \ninsert into test.test(id, type, create_time) VALUES (2, 1, toDateTime('2021-03-01 00:00:00')); \ninsert into test.test(id, type, create_time) VALUES (3, 1, toDateTime('2021-03-01 00:00:00')); \ninsert into test.test(id, type, create_time) VALUES (3, 1, toDateTime('2021-03-01 00:00:00')); \ninsert into test.test(id, type, create_time) VALUES (4, 1, toDateTime('2021-03-01 00:00:00')); \ninsert into test.test(id, type, create_time) VALUES (5, 1, toDateTime('2021-03-01 00:00:00'));\n"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"因为index_granularity设置为4，所以每4条数据就会生成一条索引记录，即使用插入的第1、5、9条数据id字段的值生成索引文件记录。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/d6\/9c\/d6ef6c6d12d669a97b0940e852caf29c.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"2.3.3 标记文件"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"mrk标记文件在primary.idx索引文件和bin数据文件之间起到了桥梁作用。primary.idx文件中的每条索引在mrk文件中都有对应的一条记录。一条记录的组成包括："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"bulletedlist","content":[{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"offset-compressed bin file"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"表示指向的压缩数据块在bin文件中的偏移量。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"offset-decompressed data block"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"表示指向的数据在解压数据块中的偏移量。"}]}]},{"type":"listitem","attrs":{"listStyle":null},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"row counts"}]}]},{"type":"listitem","attrs":{"listStyle":"none"},"content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"代表数据记录行数，小于等于index_granularity所设置的值。"}]}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/57\/d6\/57c2c50d02c4bf7637007fb3b337d6d6.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"索引，标记和数据文件下图所示："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/b2\/85\/b2dabc9da1d48abca91627a46c08dd85.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"作者：TalkingData 张凯"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"参考文档："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"1.https:\/\/clickhouse.tech\/docs"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"2.http:\/\/www.clickhouse.com.cn\/topic\/5ffec51eba8f16b55dd0ffe4"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"3.《ClickHouse原理解析与应用实践》（机械工业出版社出版，作者朱凯）"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"本文转载自公众号TalkingData（ID：Talkingdata）。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"原文链接"},{"type":"text","text":"："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"link","attrs":{"href":"https:\/\/mp.weixin.qq.com\/s\/YWSmRqOOC3F5KeDBPg2g1A","title":"","type":null},"content":[{"type":"text","text":"ClickHouse核心引擎MergeTree解读"}]}]}]}
發表評論
所有評論
還沒有人評論，想成為第一個評論的人麼? 請在上方評論欄輸入並且點擊發布.
相關文章