elasticsearch range aggs timezone 实践

在使用elasticsearch做数据存储和搜索引擎时，不可避免的会涉及到一些时间类的数据，比如：数据生成时间。本文主要针对elasticsearch使用过程中时间类型数据的mapping、写入、查询以及聚合计算等，谈谈自己的理解和总结出的最佳实践。

• Index/Type mapping

官方支持的时间类型，大致上分为两类：时间戳（支持毫秒）和日期字符串类型。考虑到可阅读性，这里我选择日期字符串类型（当然，不考虑成本的情况下，完全可以两者都使用）

• 数据写入时的时区选择

elasticsearch中相关时区的操作往往以 UTC 时间（即：0时区时间）为准，为了和 elasticsearch 保持一致，在插入数据时，我们使用 UTC 时间戳。例如：2017-12-25T16:00:00Z

• 数据筛选和聚合时的时区问题

基于上述和 elasticsearch 保持一致的 UTC 时间数据，做 range 筛选和 aggs 聚合时，考虑到数据分析脚本或者服务的本地化，我们往往使用 +08:00 的时间参数作为输入，如此在传递给elasticsearch时，由于时区不同，需要指定传入的实践参数为 +08:00 时区。

• 实例
• 构建 Index/Type Mapping

PUT http://127.0.0.1:9200/timezone
{
    "mappings": {
        "timezone-26.27": {
            "properties": {
                "@timestamp": {
                    "type": "date"
                },
                "timestring": {
                    "type": "keyword"
                }
            }
        }
    }
}

• 插入模拟数据

POST http://127.0.0.1:9200/timezone/timezone-26.27/_bulk?pretty&refresh
{"index":{"_id":1}}
{"@timestamp":"2017-12-25T16:00:00Z","timestring":"2017-12-26 00:00:00"}
{"index":{"_id":2}}
{"@timestamp":"2017-12-25T16:59:59Z","timestring":"2017-12-26 00:59:59"}
{"index":{"_id":3}}
{"@timestamp":"2017-12-25T17:00:00Z","timestring":"2017-12-26 01:00:00"}
{"index":{"_id":4}}
{"@timestamp":"2017-12-25T17:59:59Z","timestring":"2017-12-26 01:59:59"}
{"index":{"_id":5}}
{"@timestamp":"2017-12-25T18:00:00Z","timestring":"2017-12-26 02:00:00"}
{"index":{"_id":6}}
{"@timestamp":"2017-12-25T18:59:59Z","timestring":"2017-12-26 02:59:59"}
{"index":{"_id":7}}
{"@timestamp":"2017-12-25T19:00:00Z","timestring":"2017-12-26 03:00:00"}
{"index":{"_id":8}}
{"@timestamp":"2017-12-25T19:59:59Z","timestring":"2017-12-26 03:59:59"}
{"index":{"_id":9}}
{"@timestamp":"2017-12-25T20:00:00Z","timestring":"2017-12-26 04:00:00"}
{"index":{"_id":10}}
{"@timestamp":"2017-12-25T20:59:59Z","timestring":"2017-12-26 04:59:59"}
{"index":{"_id":11}}
{"@timestamp":"2017-12-25T21:00:00Z","timestring":"2017-12-26 05:00:00"}
{"index":{"_id":12}}
{"@timestamp":"2017-12-25T21:59:59Z","timestring":"2017-12-26 05:59:59"}
{"index":{"_id":13}}
{"@timestamp":"2017-12-25T22:00:00Z","timestring":"2017-12-26 06:00:00"}
{"index":{"_id":14}}
{"@timestamp":"2017-12-25T22:59:59Z","timestring":"2017-12-26 06:59:59"}
{"index&#