Shulou(Shulou.com)06/02 Report--

Http://nginxs.blog.51cto.com/

I have been exposed to ES (Elasticsearch) for more than two years since 2014, and I have a lot of feelings, especially the out-of-the-box feature of ES is completely different from the previous contact with complex hadoop and solor. ES doesn't need you to know anything about it to get started quickly, and ES's real-time search, automatic expansion, and self-healing features appeal to me. Recently, many friends have started to ask me a lot of common questions. I have summed up some of the pitfalls I have stepped on in my use in the hope that you will know more about ES.

Brief introduction

Elasticsearch is a quasi-real-time search service based on Lucene, and the search delay is in seconds. ES storage mainly parses json data by itself, and then the key in json is mapped to the fields in Lucene, and lucene is used for search and indexing. ES not only supports general full-text search and word search, but also supports fuzzy matching, synonym search, aggregation, sorting, geo and other features. ES's open source feature also makes its community active, version iteration updates quickly, now mainly divided into 2.0 and 5.0 two major versions, it is recommended that you use the latest version 5.0 will be easier to upgrade and get some more advanced features.

Here are some features that you need to know when going online or using Elasticsearch online

CPU

Es mainly depends on hard disk and memory, so the requirement for CPU is not high, generally 8 cores are fine, and if concurrency is more, it can be increased appropriately.

Hard disk

The hard disk determines the speed of your es writing and reading data. It is recommended to use a 15k mechanical hard disk and configure it as raid0, if the cluster node is H710, otherwise it is like installing a tractor engine on a ssd sports car)

Memory optimization

1. The memory usage of ES is divided into two parts: ES cache and Lucene accelerate some data through kernel cache.

2. If the server memory `G` > 64G`, the ES memory should be set to less than 32 gigabytes as far as possible, and a maximum of 31 gigabytes is recommended. Because es uses the "memory pointer compression" technology, once the memory is larger than 32 gigabytes, the technology will fail, and the effective use of memory is only 60% or 70% of the original. You don't have to worry about memory waste, because lucene caches some aggregated and sorted data through the system for quick query.

3. If the server memory `nG

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.