java - 在调用链日志存储和展现的过程中 HBase 和 ElasticSearch 如何取舍?
怪我咯
怪我咯 2017-04-18 10:33:44
0
2
550

要做微服务首先得把全链路监控搭起来,参考过一些方案,拿其中唯品会的方案来举例:

调用链信息分别存储于 HBaseElasticSearch中,ElasticSearch 单纯承担起类似于 Solr 的角色,查询的时候只需要通过 ElasticSearch 的倒排索引查到相关的 Rowkey 再去 HBase 里通过 Rowkey 查得完整调用链信息,并将起展现出来。

我知道 ElasticSearch 本身其实可以做可靠性存储的,为什么不直接将 Trace 调用链信息全量存入 ElasticSearch 中,这里借用 HBase 意义何在?

如果是为了后续的批处理计算,也感觉不妥,ES 也有 hadoop插件,可以做集成。
如果说 HBase 能够承受更大的吞吐,感觉也不对,ES 能承受的吞吐也不小。

关于这个问题我依旧思考不出个所以然来,忘各位给出见解,谢谢了!

怪我咯
怪我咯

走同样的路,发现不同的人生

reply all(2)
小葫芦

It’s not too clear how Vipshop is designed.

I know twitter is open sourcezipkin分布式追踪系统是直接将trace信息直接存储在 elasticsearch 中的。当然zipkin还支持很多其它的存储方式如mysql/nosql. zipkin

All roads lead to Rome. I don’t think you need to worry too much about why others make the choices they do, and you should pay more attention to what others have done. There are many reasons for this that outsiders cannot know. It may be personal, it may be related to the company's internal technology, or there may be other considerations.
But at least one thing is certain. To achieve the same function, you don’t have to choose hbase + elasticsearch to complete it.

巴扎黑

hbase is considered the basic version.
If you want to do analysis, you still have to use es.

Latest Downloads
More>
Web Effects
Website Source Code
Website Materials
Front End Template
About us Disclaimer Sitemap
php.cn:Public welfare online PHP training,Help PHP learners grow quickly!