blocks|key|4244622|text|它不一定以这种方式聚合值。你认为MapReduce中的映射器是如何工作的？|type|unstyled|depth|inlineStyleRanges|entityRanges|data|4244623|你可以把它想象成一个hashmap。如果它发现键已经在hashmap中，它会递增该值，如果没有，它会将键插入到hashmap中，然后初始化值。|4244624|你不能在不添加1%2B1，%2B1，%2B1，%2B1的组合的情况下添加1%2B1%2B1%2B1%2B1，除非我忘记了一些数学。|4244625|entityMap^0|0|0|0^^$0|@$1|2|3|4|5|6|7|H|8|@]|9|@]|A|$]]|$1|B|3|C|5|6|7|I|8|@]|9|@]|A|$]]|$1|D|3|E|5|6|7|J|8|@]|9|@]|A|$]]|$1|F|3|-4|5|6|7|K|8|@]|9|@]|A|$]]]|G|$]]

It doesn't necessarily aggregate values that way. How do you think a mapper in MapReduce works?

You can think of it like a hashmap. If it sees the key is already in the hashmap, it increments the value, if not, it inserts the key into the hashmap and then initializes the value.

You cant add 1+1+1+1+1 with out adding 1+1,+1,+1,+1 in some combination, unless I'm forgetting some math.

blocks|key|2182194|text|术语与Map+Reduce相同。唯一不同的是，与Map+Reduce不同的是，map+reduce只提供map和reduce函数来实现过滤、连接和聚合等逻辑，而Spark有许多函数。对于map+reduce中的map+reduce我们有spark中的map、flatMap、filter等，对于map+reduce中的reduce我们有spark中的reduce、count、reduceByKey、countByKey、aggregateByKey、groupByKey、combineByKey等。|type|unstyled|depth|inlineStyleRanges|entityRanges|data|2182195|在Map+Reduce中，如果你想在映射器端预先聚合，则需要指定组合器类，在spark组合器中，如果你使用reduceByKey，aggregateByKey，combineByKey。如果你不想使用组合器，那么你必须使用groupByKey函数。|2182196|对于您所询问的主题，map+reduce和spark中的技术术语相同。但对于分布式缓存等主题，它是不同的，在Spark中它被称为广播。此外，即使术语相同，但它们的操作方式也有细微的差异。|2182197|entityMap^0|0|0|0^^$0|@$1|2|3|4|5|6|7|H|8|@]|9|@]|A|$]]|$1|B|3|C|5|6|7|I|8|@]|9|@]|A|$]]|$1|D|3|E|5|6|7|J|8|@]|9|@]|A|$]]|$1|F|3|-4|5|6|7|K|8|@]|9|@]|A|$]]]|G|$]]

Terminology is same as Map Reduce. Only difference is unlike Map Reduce which only provides map and reduce functions to implement logic such as filtering, joins and aggregations, Spark have many out of the functions. For map in map reduce we have map, flatMap, filter etc in spark, for reduce in map reduce we have reduce, count, reduceByKey, countByKey, aggregateByKey, groupByKey, combineByKey etc in spark.

In Map Reduce you need to specify combiner class if you want to pre-aggregate on mapper side, in spark combiner is used if you use reduceByKey, aggregateByKey, combineByKey. If you do not want to use combiner, then you have to use groupByKey function. 

Technical terminology is same in map reduce and spark for the topics you have asked. But it is different for topics such as distributed cache, in Spark it is called as broadcast. Also even though terminology is same there are subtle differences the way they operate.

blocks|key|38782|text|Spark以更多的函数式方式来实现，所以函数式编程术语应该适用于这里。|type|unstyled|depth|inlineStyleRanges|entityRanges|data|38783|rdd.reduce(_+%2B+_)|code-block|syntax|javascript|38784|是reduce+(折叠/累加)函数|offset|length|style|BOLD|38785|here上有一些很好的解释，所以我只是复制粘贴其中的一些：|38786|38787|+Fold+(也称为Reduce，Accumulate)是一个高阶函数，它处理(使用组合器函数)按一定顺序排列的元素列表，并构建返回值|blockquote|38788|38789|38790|其中"%2B“是一个组合器函数|38791|38792|组合器是一个应用于两个元素的函数，并产生一个结果，该结果可以使用组合器与列表中的其余元素进行组合。|38793|38794|38795|此外，还有更多关于折叠函数及其在函数式编程Fold+(higher-order+function)中的应用的详细信息|38796|entityMap|0|LINK|mutability|MUTABLE|url|http://blogs.msdn.com/b/csliu/archive/2009/11/10/map-reduce-in-functional-programming-parallel-processing-perspectives.aspx|1|https://en.wikipedia.org/wiki/Fold_(higher-order_function)^0|0|0|1|6|F|2|0|0|4|0|0|0|0|0|0|8|5|0|0|0|0|0|L|S|1|0^^$0|@$1|2|3|4|5|6|7|1D|8|@]|9|@]|A|$]]|$1|B|3|C|5|D|7|1E|8|@]|9|@]|A|$E|F]]|$1|G|3|H|5|6|7|1F|8|@$I|1G|J|1H|K|L]|$I|1I|J|1J|K|L]]|9|@]|A|$]]|$1|M|3|N|5|6|7|1K|8|@]|9|@$I|1L|J|1M|1|1N]]|A|$]]|$1|O|3|-4|5|6|7|1O|8|@]|9|@]|A|$]]|$1|P|3|Q|5|R|7|1P|8|@]|9|@]|A|$]]|$1|S|3|-4|5|6|7|1Q|8|@]|9|@]|A|$]]|$1|T|3|-4|5|6|7|1R|8|@]|9|@]|A|$]]|$1|U|3|V|5|6|7|1S|8|@$I|1T|J|1U|K|L]]|9|@]|A|$]]|$1|W|3|-4|5|6|7|1V|8|@]|9|@]|A|$]]|$1|X|3|Y|5|R|7|1W|8|@]|9|@]|A|$]]|$1|Z|3|-4|5|6|7|1X|8|@]|9|@]|A|$]]|$1|10|3|-4|5|6|7|1Y|8|@]|9|@]|A|$]]|$1|11|3|12|5|6|7|1Z|8|@]|9|@$I|20|J|21|1|22]]|A|$]]|$1|13|3|-4|5|6|7|23|8|@]|9|@]|A|$]]]|14|$15|$5|16|17|18|A|$19|1A]]|1B|$5|16|17|18|A|$19|1C]]]]

Spark does it in more functional way, so functional programming terminology should be applicable here.

<pre><code>rdd.reduce(_ + _)
</code></pre>

is a reduce (fold/accumulate) function:

There are some good explanations available <a href="http://blogs.msdn.com/b/csliu/archive/2009/11/10/map-reduce-in-functional-programming-parallel-processing-perspectives.aspx" rel="nofollow" title="here">here</a>, so I'm just copy-pasting some of them:

<blockquote>
 Fold (a.k.a. Reduce, Accumulate) is a higher order function that processes (using a combiner function) a list of elements in some order and build up a return value
</blockquote>

Where "+" here is a combiner function:

<blockquote>
 Combiner is a function that is applied to two elements and produces a result that can be combined using combiner with the remaining elements in the list.
</blockquote>

Also there are more details available on fold functions in general and their application in functional programming <a href="https://en.wikipedia.org/wiki/Fold_(higher-order_function)" rel="nofollow">Fold (higher-order function)</a>

In a word count example below:

<pre><code>(Hello,1)
(Hello,1)
(Hello,1)
(Hello,1)
(Hello,1)
</code></pre>

The reducer function of Hadoop would collect all 5 elements with key "Hello", and then do the aggregation in the reducer function.

However, in Spark, it is actually reduces 2 elements each time. For example, it combines first and second (Hello,1) to (Hello,2), combines the third and forth (Hello,1) to (Hello,2) ... etc (of course, the real case might be in different order).

So is there a specific terminology to describe such computational method used in Spark? Thanks!

Spark: Technical terminology for reduce elements on the run-time?

翻译质量差，导致语言生硬或混乱。

没有提供实际的解决方法或示例。

解答不清晰，无法理解或解决问题。

页面排版不美观，阅读体验差。

文章

问答

视频

学习中心

腾讯云实验室

直播

竞赛

腾讯云代码分析专区

腾讯iOA零信任安全管理系统专区

腾讯云架构师技术同盟交流圈

腾讯云数据库专区

腾讯云顾问专区

腾讯云原生专区

腾讯混元专区

腾讯云TCE专区

腾讯云Lighthouse专区

腾讯云HAI专区

腾讯云Edgeone专区

腾讯云存储专区

腾讯云智能专区

腾讯轻联专区 

腾讯云开发专区

TAPD专区

腾讯轻量云游戏服专区

腾讯云最具价值专家

腾讯云架构师技术同盟

腾讯云创作之星

腾讯云开发者先锋

腾讯云代码助手

云原生构建

TAPD 敏捷项目管理

Cloud Studio

SDK中心

API中心

命令行工具

涵盖代码开发、场景应用、自动测试全流程，助你从零构建专属AI助手

一站式MCP教程库，解锁AI应用新玩法

在下面的字数统计示例中：(Hello,1)(Hello,1)(Hello,1)(Hello,1)(Hello,1)Hadoop的reducer函数将使用键"Hello“收集所有5个元素，然后在reducer函数中进行聚合。然而，在Spark中，它实际上是每次减少2个元素。例如，它将第一个和第二个(Hello,1)组合为...

问Spark:运行时reduce元素的技术术语？
EN

Stack Overflow用户

社区

活动

圈层

关于

腾讯云开发者

热门产品

热门推荐

更多推荐

问Spark:运行时reduce元素的技术术语？EN