Presto + Druid : How Big Data Goes Real Time @ Twitter

案例背景:
An important characteristic of Twitter is its real-time nature. Consequently, many of Twitter’s projects need real time analytics as a platform service. During recent years, a number of teams are adopting Presto and Druid as real time analytics engines.

解決思路:
In this talk, We start with Twitter’s big data architecture, followed by a detail introduction of Presto and Druid. We will focus on our design and implementation of Presto Druid Connector, which provides real time query latency with full SQL functionality. Leveraging Presto Connectors, users are able to join data between a variety of data sources, without data copy, including Hadoop, Druid, MySQL, Elasticsearch, etc. With predicate and aggregation pushdown, we achieved sub second query latency with Presto Druid Connector. We will also share our production experience, and roadmaps.

成果：
通過開源項(xiàng)目Presto和Druid的研發(fā)，我們實(shí)現(xiàn)了大數(shù)據(jù)分析的實(shí)時(shí)處理，絕大多數(shù)的查詢可以做到一秒之內(nèi)完成，很好的支持了Twitter業(yè)務(wù)