One to two years ahead of MongoDB. Build Enterprise-Class NoSQL database

Over the past few years, NoSQL database with its easy to expand, high-performance, high availability, data model flexibility and other features attracted a large number of emerging internet companies, including the domestic Taobao, Sina, Jingdong Mall, Sogou, etc. have been in the local attempt to NoSQL solutions. Guangzhou giant FIR Database is a focus on the new NoSQL distributed database research and development of the start-up company, has now won the first round of angel investment, the core product is sequoiadb. The founding team members of the giant FIR database are from IBM North American laboratories and have long been engaged in research and development of relational database DB2. SEQUOIADB 1.3 was officially released in April 2013 (the latest version is 1.5), mainly for the government, telecommunications, finance, electricity and the internet and other industries with a large number of business data solutions, and its customers include more than 500 companies in the world, such as domestic well-known banks, telecommunications and internet enterprises.

More focused on the enterprise-level market

The biggest difference between sequoiadb and the current mainstream NoSQL database is that it focuses more on the enterprise market and develops many features. The Giant FIR database co-founder/cto Wang Tao is a former IBM DB2 North American Laboratory Senior Research and Development member, one of the world's top advisory team members (the world only 15 people). In his view, there are many excellent nosql products in the market at present, such as Redis, MongoDB, HBase and so on, each product has its own characteristics, but it has its obvious defects. For example, SEQUOIADB is a document class database, and MongoDB are categorized as belonging to the same type, but the two sides walk in a different way. MongoDB born with the internet gene, its function and demand is based on the Internet model to create, so in the application of the enterprise-level market, the inevitable will appear acclimatized to the scene.

Reporter in and Minsheng Bank science and technology Yuan Spring communication learned that, as the first batch of domestic banking involved in large data technology Minsheng Bank, in the use of existing NoSQL products (such as MongoDB) in the process of discovering its lack of many enterprise-class features, such as lack of transaction and SQL support, Integration with Hadoop systems, especially MongoDB, is unfriendly to reusing existing investments, including in terms of continuity of developer skills, and docking of existing applications. These for the traditional enterprises such as Minsheng Bank, the information island brought about by the architectural damage is fatal, it is far greater than the value of this product.

By introducing the NoSQL product sequoiadb, such as transaction, SQL support, data compression, integration with Hadoop systems, the ability to run on IBM power machines to meet the needs of enterprise-class users, and make existing skills and investments as reusable as possible. In particular, SEQUOIADB provides the SQL interface, and the traditional relational database docking functions are mongodb at this stage can not be provided, not even planned to provide. MongoDB, on the other hand, is also accelerating penetration into industry users, especially in North America, "said Wang Tao," even if it is to catch up with SEQUOIADB's core technological advantage in the enterprise-level sector, MongoDB is still a year or two away.

In addition, it has many unique features for other databases, such as HBase, but still fails to meet the needs of enterprise users, such as integration with Hadoop systems. For example, a bank business scenario, the bank in the transaction flow after archiving, need to be based on a number of different field conditions for fast real-time query. SEQUOIADB can search for the required data from dozens of terabytes of data in milliseconds to seconds, using features such as partitioned tables, multiple indexes, and so on. With HBase, because only the primary key index is supported, it takes a few 10 minutes to scan all the data when querying through a non-key field, which is not enough to meet the demand.

Open source database engine and code not used

It seems that the traditional database has a lot of unsuitable scenarios and technical limitations, the main limitations of the data model is rigid, poor scalability, processing of massive data performance bottlenecks, lack of processing semi-structured and unstructured data. In recent years, with the large number of unstructured and semi-structured data entering the enterprise, relational database has become more and more limited and insufficient. Wang Tao thinks, "The appearance of NoSQL database can be used to make up these flaw, but NoSQL database lacks enterprise level gene and commercial grade technical service support, this is also the reason that we do sequoiadb." ”

In addition, the core products SEQUOIADB by former IBM DB2 Senior research and development team in North America to complete the prototype design and kernel development. The development process did not use open source database engine and code, this is because the current mainstream of the NoSQL database program is based on the GPL is basically open source, which is not in line with the strategy of business-level database. But it doesn't rule out the possibility of sequoiadb open source in the future at the right time. At this stage, the most headaches for Wang Tao is that it is difficult to find a technical talent for Hadoop distributed and NoSQL database technology, or has been dug by some giants, even through such as technical open classes and many other ways, can find the right talent is still limited. The training of technical personnel will be their next priority.

A year ago, there were a large number of industry users looking at the NoSQL database solution, but this year, more and more enterprise users have begun to do POC (Proof of concept, concept validation), and began to deploy the corresponding solution. While NoSQL will not quickly replace SQL, NoSQL provides features that are needed for new age applications. Wang Tao believes that relational and non relational databases are not substituted relationships, but that they will coexistence, with over 70% of existing relational database scenarios being replaced by NoSQL.