Abstract: HBase is a database and a storage. HBase with dual attributes has a wide range of application scenarios. In 2.0, OffHeap was introduced to reduce latency and meet online requirements. By introducing MOB, objects of around 10M can be stored, which is completely adapted to object storage. It is also the most competitive engine due to its concurrency and storage capabilities
Summary of HBase
HBase is a distributed storage and database engine that supports tens of millions of QPS and PB storage devices. These devices have been verified in production environments and in many companies. In particular, Alibaba, Xiaomi, JD.com, and Didi all have tens of thousands of HBase clusters. The first requirement for choosing a technology is to align with large companies, which invest a lot of people to maintain, improve, and contribute to the community.
The relationship between NewSQL and NoSQL
Technology is always moving forward, and NewSQL is now being talked about, but NewSQL in my opinion is an encapsulation, a sub-scenario on top of NoSQL. Large tables in NoSQL typically provide KV1V2... Vn, where each V can be 1b or 100MB. It's kind of a meta-existence, like 01 in the digital world, which can be combined in any way. In NoSQL represented by HBase, HBase can combine any scenario. NewSQL can be a sub-scenario in which an SQL layer is added or transactions are added at a later layer.
About computing and storage separation
In the cloud, the core of the engine is storage computing separation, storage can be charged on demand, at least elastic scaling. Computing is provided on a node storage basis and charged entirely on QPS, which is either prohibitively expensive or difficult to accommodate for more scenarios. For example, if you store 10 megabytes, how many QPS does it take? Because HBase is inherently separated from storage and computing, it naturally ADAPTS to cloud architectures. In terms of cloud architectures, HBase has advantages.
HBase is a database or a storage device. HBase with dual attributes has a wide range of application scenarios. In 2.0, OffHeap was introduced to reduce latency and meet online requirements. By introducing MOB, objects of around 10M can be stored, which is completely adapted to object storage. It is also the most competitive engine due to its concurrency and storage capabilities
- Object storage: Many headlines, news, web pages, and pictures are stored in HBase. Some virus databases of virus companies are also stored in HBase
- Timing data: HBase has the OpenTSDB module, which meets the requirements of timing scenarios
- Recommended portrait: The user portrait, in particular, is a large sparse matrix. Risk control of ants is based on HBase
- Spatial-temporal data: mainly tracks and weather grids. Track data of Didi Taxi is mainly stored in HBase. In addition, data of Internet of vehicles enterprises with larger data volume are stored in HBase
- CubeDB OLAP: Kylin is a cube analysis tool. The underlying data is stored in HBase. Many customers build cubes based on offline computing and store them in HBase to meet the requirements of online report query
- Message/order: In the telecom and banking sectors, many order query applications are stored at the bottom, and many communication and message synchronization applications are built on HBase
- Feeds stream: The typical application is xx moments similar applications
- NewSQL: The Phoenix plug-in can meet the requirements of secondary index, SQL, and non-transactional SQL for connecting traditional data
To read more articles, please scan the following QR code: