Real-time synchronization of-hbase data to elasticsearch for high stability and high performance (one)

Recently has been focus in the construction of ETL data Center this piece, the need to put hbase several tables of data in real-time synchronization to Elasticsearch, research for a period of time, on the Internet can refer to just a few documents: 1. The HBase data is synchronized to the Elasticsearch scenario 2. Synchronize data to Elasticsearch 3 using the HBase collaborator (Coprocessor). HBase official documentation. 4. ElasticSearch Java API finally choose to use the coprocessor implementation, according to the second information can quickly make a running demo, at this time ecstatic, unexpectedly this is the beginning of Long March.
In a word, there are many problems: how to support multiple tables to share a single synchronization jar package. How to continuously and stably interact with ES, write data. How to minimize the risk of downtime for hbase caused by Coprocessor's intrusive nature. How to achieve synchronization efficiently. In order to solve these problems one by one, ate a lot of hardships, divided into three articles to share to everyone.