开头还是介绍一下群，如果感兴趣PolarDB ,MongoDB ,MySQL ,PostgreSQL ,Redis, Oceanbase, Sql Server等有问题，有需求都可以加群群内有各大数据库行业大咖，CTO，可以解决你的问题。加群请联系 liuaustin3 ，（共2280人左右 1 + 2 + 3 + 4 +5） 新人奖直接分配到5群,5群超过300 已建立6群。
上期我们针对MongoDB的聚合操作进行了一个实例的操作并且发现了与传统数据库在操作和索引方面的有意思的不同。这次我们来继续深入聚合操作，我们这里换一个数据集合collection ,将我们的复杂度提高。
（上期：MongoDB 挑战传统数据库聚合查询，干不死他们的）
mongo7 [direct: primary] test> show collections;
test
mongo7 [direct: primary] test> db.test.find();
[
{
_id: '01001',
city: 'AGAWAM',
loc: [ -72.622739, 42.070206 ],
pop: 15338,
state: 'MA'
},
{
_id: '01002',
city: 'CUSHMAN',
loc: [ -72.51565, 42.377017 ],
pop: 36963,
state: 'MA'
},
{
_id: '01007',
city: 'BELCHERTOWN',
loc: [ -72.410953, 42.275103 ],
pop: 10579,
state: 'MA'
},这个collection 是一个记录城市的经纬度的集合，里面有城市和州的名字以及具体的经纬度等信息。下面我们要通过几个案例来说明，到底聚合该怎么去撰写，与传统的数据库有多大的不同。问题1 ：以上数据中，针对洲名相同城市名相同，重复出现的次数，这些重复出现的次数的总和是多少？
mongo7 [direct: primary] test> db.test.aggregate([
... {
... $group: {
... _id: { state: "$state", city: "$city" },
... count: { $sum: 1 }
... }
... },
... {
... $match: {
... count: { $gt: 1 }
... }
... },
... {
... $count: "duplicateCityStateCount"
... }
... ])
[ { duplicateCityStateCount: 893 } ]
Enterprise mongo7 [direct: primary] test>我们先看第一个列子，这个例子中，我们是以state,city作为分组的对象，然后对于每个分组的对象进行计数，然后发现其中超过1 次的技术对象进行数据的过滤，最终我们计算出到底有多少state city 在数据中出现的次数超过2次以上的总体出现的次数。但如果将这个语句换成SQL 则比较难来实现，但下面的例子，SQL会比较容易实现，如
db.test.aggregate([
{
$group: {
_id: { state: "$state", city: "$city" }, // 按州和城市进行分组
count: { $sum: 1 } // 计算每个组中文档的数量
}
},
{
$sort: { count: -1 } // 按照文档数量降序排序
}
])SELECT state, city, COUNT(*) AS count
FROM test
GROUP BY state, city
HAVING COUNT(*) > 1;上面的SQL 语句和MONGODB 的语句表达的意思是一致的，意思是针对每个城市和州，重复出现的次数的分组统计
Enterprise mongo7 [direct: primary] test> db.test.aggregate([
... {
... $group: {
... _id: { state: "$state", city: "$city" }, // 按州和城市进行分组
... count: { $sum: 1 } // 计算每个组中文档的数量
... }
... },
... {
... $sort: { count: -1 } // 按照文档数量降序排序
... }
... ])
[
{ _id: { state: 'TX', city: 'HOUSTON' }, count: 93 },
{ _id: { state: 'CA', city: 'LOS ANGELES' }, count: 56 },
{ _id: { state: 'PA', city: 'PHILADELPHIA' }, count: 48 },
{ _id: { state: 'IL', city: 'CHICAGO' }, count: 47 },
{ _id: { state: 'TX', city: 'SAN ANTONIO' }, count: 45 },
{ _id: { state: 'TX', city: 'DALLAS' }, count: 44 },
{ _id: { state: 'MO', city: 'KANSAS CITY' }, count: 41 },
{ _id: { state: 'NY', city: 'NEW YORK' }, count: 40 },
{ _id: { state: 'TX', city: 'AUSTIN' }, count: 40 },
{ _id: { state: 'NY', city: 'BROOKLYN' }, count: 37 },
{ _id: { state: 'CA', city: 'SAN DIEGO' }, count: 34 },
{ _id: { state: 'FL', city: 'MIAMI' }, count: 34 },
{ _id: { state: 'OK', city: 'OKLAHOMA CITY' }, count: 33 },
{ _id: { state: 'AZ', city: 'PHOENIX' }, count: 33 },
{ _id: { state: 'GA', city: 'ATLANTA' }, count: 31 },
{ _id: { state: 'CA', city: 'SAN JOSE' }, count: 29 },
{ _id: { state: 'CA', city: 'SACRAMENTO' }, count: 28 },
{ _id: { state: 'OK', city: 'TULSA' }, count: 28 },
{ _id: { state: 'NE', city: 'OMAHA' }, count: 27 },
{ _id: { state: 'CA', city: 'SAN FRANCISCO' }, count: 26 }
]这里我们将查询计划进行答应看看，这个语句到底是怎么执行的
Enterprise mongo7 [direct: primary] test> db.test.aggregate([
... {
... $group: {
... _id: { state: "$state", city: "$city" },
... count: { $sum: 1 }
... }
... },
... {
... $sort: { count: -1 }
... }
... ]).explain("executionStats")
{
explainVersion: '2',
stages: [
{
'$cursor': {
queryPlanner: {
namespace: 'test.test',
indexFilterSet: false,
parsedQuery: {},
queryHash: '28FD9B9E',
planCacheKey: 'B81730B0',
maxIndexedOrSolutionsReached: false,
maxIndexedAndSolutionsReached: false,
maxScansToExplodeReached: false,
winningPlan: {
queryPlan: {
stage: 'GROUP',
planNodeId: 2,
inputStage: {
stage: 'COLLSCAN',
planNodeId: 1,
filter: {},
direction: 'forward'
}
},
slotBasedPlan: {
slots: '$$RESULT=s12 env: { s3 = Timestamp(1714570591, 1) (CLUSTER_TIME), s4 = 1714570597137 (NOW), s2 = Nothing (SEARCH_META), s1 = TimeZoneDatabase(Turkey...Asia/Chungking) (timeZoneDB) }',
stages: '[2] mkobj s12 [_id = s11, count = s9] true false \n' +
'[2] project [s11 = newObj("state", s6, "city", s5)] \n' +
'[2] group [s6, s5] [s9 = sum(1)] spillSlots[s10] mergingExprs[sum(s10)] \n' +
'[1] scan s7 s8 none none none none lowPriority [s5 = city, s6 = state] @"4a56b246-67a5-4b06-b71b-0d8ec25876c2" true false '
}
},
rejectedPlans: []
},
executionStats: {
executionSuccess: true,
nReturned: 25701,
executionTimeMillis: 168,
totalKeysExamined: 0,
totalDocsExamined: 29353,
executionStages: {
stage: 'mkobj',
planNodeId: 2,
nReturned: 25701,
executionTimeMillisEstimate: 98,
opens: 1,
closes: 1,
saveState: 33,
restoreState: 33,
isEOF: 1,
objSlot: 12,
fields: [],
projectFields: [ '_id', 'count' ],
projectSlots: [ Long("11"), Long("9") ],
forceNewObject: true,
returnOldObject: false,
inputStage: {
stage: 'project',
planNodeId: 2,
nReturned: 25701,
executionTimeMillisEstimate: 63,
opens: 1,
closes: 1,
saveState: 33,
restoreState: 33,
isEOF: 1,
projections: { '11': 'newObj("state", s6, "city", s5) ' },
inputStage: {
stage: 'group',
planNodeId: 2,
nReturned: 25701,
executionTimeMillisEstimate: 41,
opens: 1,
closes: 1,
saveState: 33,
restoreState: 33,
isEOF: 1,
groupBySlots: [ Long("6"), Long("5") ],
expressions: { '9': 'sum(1) ', initExprs: { '9': null } },
mergingExprs: { '10': 'sum(s10) ' },
usedDisk: false,
spills: 0,
spilledRecords: 0,
spilledDataStorageSize: 0,
inputStage: {
stage: 'scan',
planNodeId: 1,
nReturned: 29353,
executionTimeMillisEstimate: 40,
opens: 1,
closes: 1,
saveState: 33,
restoreState: 33,
isEOF: 1,
numReads: 29353,
recordSlot: 7,
recordIdSlot: 8,
fields: [ 'city', 'state' ],
outputSlots: [ Long("5"), Long("6") ]
}
}
}
}
}
},
nReturned: Long("25701"),
executionTimeMillisEstimate: Long("139")
},
{
'$sort': { sortKey: { count: -1 } },
totalDataSizeSortedBytesEstimate: Long("12869626"),
usedDisk: false,
spills: Long("0"),
spilledDataStorageSize: Long("0"),
nReturned: Long("25701"),
executionTimeMillisEstimate: Long("147")
}
],
serverInfo: {
host: 'mongo',
port: 27017,
version: '7.0.1',
gitVersion: '425a0454d12f2664f9e31002bbe4a386a25345b5'
},
serverParameters: {
internalQueryFacetBufferSizeBytes: 104857600,
internalQueryFacetMaxOutputDocSizeBytes: 104857600,
internalLookupStageIntermediateDocumentMaxSizeBytes: 104857600,
internalDocumentSourceGroupMaxMemoryBytes: 104857600,
internalQueryMaxBlockingSortMemoryUsageBytes: 104857600,
internalQueryProhibitBlockingMergeOnMongoS: 0,
internalQueryMaxAddToSetBytes: 104857600,
internalDocumentSourceSetWindowFieldsMaxMemoryBytes: 104857600,
internalQueryFrameworkControl: 'trySbeEngine'
},
command: {
aggregate: 'test',
pipeline: [
{
'$group': {
_id: { state: '$state', city: '$city' },
count: { '$sum': 1 }
}
},
{ '$sort': { count: -1 } }
],
cursor: {},
'$db': 'test'
},
ok: 1,
'$clusterTime': {
clusterTime: Timestamp({ t: 1714570591, i: 1 }),
signature: {
hash: Binary(Buffer.from("7a5832f6c1d7fc8b6b9bfe23a274a5b4f417cdcd", "hex"), 0),
keyId: Long("7364003857451450369")
}
},
operationTime: Timestamp({ t: 1714570591, i: 1 })这里我们在提出一个新的需求，值对州为德克萨斯的城市进行类似的数据统计。同时针对这个查询我们仅仅这对state 添加索引就可以进行查询效率的简单提升。下面的执行计划已经明显的开始使用IXSCAN
Enterprise mongo7 [direct: primary] test> db.test.aggregate([ { $match: { state: "TX" } }, { $group: { _id: { state: "$state", city: "$city" }, count: { $sum: 1 } } }, { $match: { count: { $gt: 1 } } }] ).explain("executionStats")
{
explainVersion: '2',
stages: [
{
'$cursor': {
queryPlanner: {
namespace: 'test.test',
indexFilterSet: false,
parsedQuery: { state: { '$eq': 'TX' } },
queryHash: '57E70BDE',
planCacheKey: 'FC027C07',
maxIndexedOrSolutionsReached: false,
maxIndexedAndSolutionsReached: false,
maxScansToExplodeReached: false,
winningPlan: {
queryPlan: {
stage: 'GROUP',
planNodeId: 3,
inputStage: {
stage: 'FETCH',
planNodeId: 2,
inputStage: {
stage: 'IXSCAN',
planNodeId: 1,
keyPattern: { state: 1 },
indexName: 'state_1',
isMultiKey: false,
multiKeyPaths: { state: [] },
isUnique: false,
isSparse: false,
isPartial: false,
indexVersion: 2,
direction: 'forward',
indexBounds: { state: [ '["TX", "TX"]' ] }
}
}
},
slotBasedPlan: {
slots: '$$RESULT=s19 env: { s3 = Timestamp(1714571396, 7) (CLUSTER_TIME), s1 = TimeZoneDatabase(Turkey...Asia/Chungking) (timeZoneDB), s2 = Nothing (SEARCH_META), s4 = 1714571398752 (NOW), s6 = KS(3C5458000104), s7 = KS(3C545800FE04), s11 = {"state" : 1} }',
stages: '[3] mkobj s19 [_id = s18, count = s16] true false \n' +
'[3] project [s18 = newObj("state", s15, "city", s14)] \n' +
'[3] group [s15, s14] [s16 = sum(1)] spillSlots[s17] mergingExprs[sum(s17)] \n' +
'[2] nlj inner [] [s5, s8, s9, s10, s11] \n' +
' left \n' +
' [1] cfilter {(exists(s6) && exists(s7))} \n' +
' [1] ixseek s6 s7 s10 s5 s8 s9 [] @"4a56b246-67a5-4b06-b71b-0d8ec25876c2" @"state_1" true \n' +
' right \n' +
' [2] limit 1 \n' +
' [2] seek s5 s12 s13 s8 s9 s10 s11 [s14 = city, s15 = state] @"4a56b246-67a5-4b06-b71b-0d8ec25876c2" true false \n'
}
},
rejectedPlans: []
},
executionStats: {
executionSuccess: true,
nReturned: 1233,
executionTimeMillis: 12,
totalKeysExamined: 1671,
totalDocsExamined: 1671,
executionStages: {
stage: 'mkobj',
planNodeId: 3,
nReturned: 1233,
executionTimeMillisEstimate: 6,
opens: 1,
closes: 1,
saveState: 2,
restoreState: 2,
isEOF: 1,
objSlot: 19,
fields: [],
projectFields: [ '_id', 'count' ],
projectSlots: [ Long("18"), Long("16") ],
forceNewObject: true,
returnOldObject: false,
inputStage: {
stage: 'project',
planNodeId: 3,
nReturned: 1233,
executionTimeMillisEstimate: 6,
opens: 1,
closes: 1,
saveState: 2,
restoreState: 2,
isEOF: 1,
projections: { '18': 'newObj("state", s15, "city", s14) ' },
inputStage: {
stage: 'group',
planNodeId: 3,
nReturned: 1233,
executionTimeMillisEstimate: 6,
opens: 1,
closes: 1,
saveState: 2,
restoreState: 2,
isEOF: 1,
groupBySlots: [ Long("15"), Long("14") ],
expressions: { '16': 'sum(1) ', initExprs: { '16': null } },
mergingExprs: { '17': 'sum(s17) ' },
usedDisk: false,
spills: 0,
spilledRecords: 0,
spilledDataStorageSize: 0,
inputStage: {
stage: 'nlj',
planNodeId: 2,
nReturned: 1671,
executionTimeMillisEstimate: 6,
opens: 1,
closes: 1,
saveState: 2,
restoreState: 2,
isEOF: 1,
totalDocsExamined: 1671,
totalKeysExamined: 1671,
collectionScans: 0,
collectionSeeks: 1671,
indexScans: 0,
indexSeeks: 1,
indexesUsed: [ 'state_1' ],
innerOpens: 1671,
innerCloses: 1,
outerProjects: [],
outerCorrelated: [
Long("5"),
Long("8"),
Long("9"),
Long("10"),
Long("11")
],
outerStage: {
stage: 'cfilter',
planNodeId: 1,
nReturned: 1671,
executionTimeMillisEstimate: 6,
opens: 1,
closes: 1,
saveState: 2,
restoreState: 2,
isEOF: 1,
numTested: 1,
filter: '(exists(s6) && exists(s7)) ',
inputStage: {
stage: 'ixseek',
planNodeId: 1,
nReturned: 1671,
executionTimeMillisEstimate: 6,
opens: 1,
closes: 1,
saveState: 2,
restoreState: 2,
isEOF: 1,
indexName: 'state_1',
keysExamined: 1671,
seeks: 1,
numReads: 1672,
indexKeySlot: 10,
recordIdSlot: 5,
snapshotIdSlot: 8,
indexIdentSlot: 9,
outputSlots: [],
indexKeysToInclude: '00000000000000000000000000000000',
seekKeyLow: 's6 ',
seekKeyHigh: 's7 '
}
},
innerStage: {
stage: 'limit',
planNodeId: 2,
nReturned: 1671,
executionTimeMillisEstimate: 0,
opens: 1671,
closes: 1,
saveState: 2,
restoreState: 2,
isEOF: 1,
limit: 1,
inputStage: {
stage: 'seek',
planNodeId: 2,
nReturned: 1671,
executionTimeMillisEstimate: 0,
opens: 1671,
closes: 1,
saveState: 2,
restoreState: 2,
isEOF: 0,
numReads: 1671,
recordSlot: 12,
recordIdSlot: 13,
seekKeySlot: 5,
snapshotIdSlot: 8,
indexIdentSlot: 9,
indexKeySlot: 10,
indexKeyPatternSlot: 11,
fields: [ 'city', 'state' ],
outputSlots: [ Long("14"), Long("15") ]
}
}
}
}
}
}
}
},
nReturned: Long("1233"),
executionTimeMillisEstimate: Long("6")
},
{
'$match': { count: { '$gt': 1 } },
nReturned: Long("67"),
executionTimeMillisEstimate: Long("6")
}
],
serverInfo: {
host: 'mongo',
port: 27017,
version: '7.0.1',
gitVersion: '425a0454d12f2664f9e31002bbe4a386a25345b5'
},
serverParameters: {
internalQueryFacetBufferSizeBytes: 104857600,
internalQueryFacetMaxOutputDocSizeBytes: 104857600,
internalLookupStageIntermediateDocumentMaxSizeBytes: 104857600,
internalDocumentSourceGroupMaxMemoryBytes: 104857600,
internalQueryMaxBlockingSortMemoryUsageBytes: 104857600,
internalQueryProhibitBlockingMergeOnMongoS: 0,
internalQueryMaxAddToSetBytes: 104857600,
internalDocumentSourceSetWindowFieldsMaxMemoryBytes: 104857600,
internalQueryFrameworkControl: 'trySbeEngine'
},
command: {
aggregate: 'test',
pipeline: [
{ '$match': { state: 'TX' } },
{
'$group': {
_id: { state: '$state', city: '$city' },
count: { '$sum': 1 }
}
},
{ '$match': { count: { '$gt': 1 } } }
],
cursor: {},
'$db': 'test'
},
ok: 1,
'$clusterTime': {
clusterTime: Timestamp({ t: 1714571396, i: 7 }),
signature: {
hash: Binary(Buffer.from("b50924fe412dcbca2ee2bc00a6592d26c5c39d52", "hex"), 0),
keyId: Long("7364003857451450369")
}
},
operationTime: Timestamp({ t: 1714571396, i: 7 })
}
Enterprise mongo7 [direct: primary] test>如果将上面的MOGNODB 的语句翻译成SQL 可以翻译成
SELECT state, city, COUNT(*) AS count
FROM collection
WHERE state = 'TX'
GROUP BY state, city
HAVING COUNT(*) > 1;写到这里我们，我们回顾一下，在MOGNODB 的数据处理里面，有一些写法，的确无法直接翻译成SQL语句，或者SQL语句通过简单的写法无法直接表达，并且我们也应该熟知，在mongodb的数据处理中，也可以通过分部的方式来处理，比如，不一次这对以state 为完全分组的方式来统计city 的数据，可以针对state 建立索引，并逐个对于以state为条件的方式city的重复数进行统计，针对这样的方式也可以先将state作为固定的输入，通过javascript + 查询语句的方式来处理。
结论，Mongodb的查询语句要比SQL语句更灵活，方案更多，优化的点更多，非常适合程序员来通过Mongodb 来继续数据的统计分析。
置顶文章：
临时工说：炮轰阿里云MongoDB司令部 低质高价技术差 你是要疯！！！！
临时工访谈：NoSQL 大有前景，MongoDB DBA 被裁员后谋求新职位
MongoDB 有那么难吗? 你死不死 ！ (语言粗暴，心里脆弱别看)
往期热门文章：
MySQL 8.0 小版本更新要点，那个小版本更稳定（8.0.24-8.0.37）
PostgreSQL 哪些版本尽量避免使用，版本更新重点明晰（PG12）
SQL SERVER 2022 针对缓存扫描和Query Store 的进步，可以考虑进行版本升级
临时工访谈：腾讯“退休”的架构师怎么看数据库 和 DBA 在项目中的重要性
临时工访谈：无名氏意外到访-- 也祝你好运（管理者PUA DBA现场直播）
临时工说：经济规律解读ORACLE 工资低 --读 Roger 数据库专栏
PostgreSQL 为什么也不建议 RR隔离级别，MySQL别笑
临时工访谈：OceanBase上海开大会，我们四个开小会 OB 国产数据库破局者
临时工说：OceanBase 到访，果然数据库的世界很卷，没边
PolarDB for PostgreSQL 有意思吗？有意思呀
PostgreSQL 玩PG我们是认真的，vacuum 稳定性平台我们有了
临时工说：裁员裁到 DBA 咋办 临时工教你 套路1 2 3
临时工说：OceanBase 到访，果然数据库的世界很卷，没边
MONGODB ---- Austindatabases 历年文章合集
MYSQL --Austindatabases 历年文章合集
POSTGRESQL --Austindatabaes 历年文章整理
POLARDB -- Ausitndatabases 历年的文章集合
PostgreSQL 查询语句开发写不好是必然，不是PG的锅
SQL SERVER 如何实现UNDO REDO 和PostgreSQL 有近亲关系吗
MongoDB 2023纽约 MongoDB 大会 -- 我们怎么做的新一代引擎 SBE Mongodb 7.0双擎力量（译）
MongoDB 2023年度纽约 MongoDB 年度大会话题 -- MongoDB 数据模式与建模
MongoDB 会丢数据吗？在次补刀MongoDB 双机热备
临时工说：从人性的角度来分析为什么公司内MySQL 成为少数派，PolarDB 占领高处
PostgreSQL 字符集乌龙导致数据查询排序的问题，与 MySQL 稳定 "PG不稳定"
PostgreSQL Patroni 3.0 新功能规划 2023年 纽约PG 大会 （音译）
Austindatabases 公众号，主要围绕数据库技术（PostgreSQL, MySQL, Mongodb, Redis, SqlSERVER，PolarDB, Oceanbase 等）和职业发展，国外数据库大会音译,国外大型IT信息类网站文章翻译，等，希望能和您共同发展。
扫一扫
8万+
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
