Nosql--mongodb selection

NoSQL databases are, of course, an essential part of the past architecture selection and practical use. We used the typical NoSQL databases of cassandra,mongodb,redis,hbase, respectively.

1,cassandra:

Cassandra from the installation configuration, to the use, load balancing mechanism, and so on, is undoubtedly one of the most convenient uses of these emerging nosql (personal use experience view)

But from recent news, due to a number of more serious database stop service incidents, Cassandra founder of Facebook, and Twitter began to gradually abandon the Cassandra, only Cassandra used in non-core modules, not Digg is still in use, It seems that we should treat it with caution. 2008 Facebook has let Cassandra Open source to Apache.

2.MongoDB:

Its style can be said, in today's Webapi popular era, it is more easy to use, Bjson operation style, automatic data balance mechanism (of course to beware of storage debris problem), relative MySQL and other SQL database has excellent consideration comprehensive, distributed scheme, automatic m/s master-slave read/write switch. For the data cluster, it can be said that the very perfect sharding and other automation support. The most serious event that has ever been heard is Foursquare's 11-hour database outage event. Relatively acceptable:), it is written using c++/boost, efficiency performance is indeed good.

3.Redis:

It's an efficient in-memory database that uses it to persist data storage, which is nonsense, and if it's actually used like any other nosql (considering read-write consistency or write security), it slows down:) But he offers more operational data types than memcached, It can be used entirely as an efficient and easy-to-use cache, Benchmark is said to be better than memcached. The size of the data I use is not so big, I dare not comment.

4.HBase:

Conceptually also relatively perfect, with the support of the Hive Open Source Tool, which makes hbase easier to use relative to other NoSQL databases, based on the HDFs Distributed File system, makes hbase inherently well supported for mass distribution clusters. And because of the concomitant with Hadoop, so a system to use data analysis, intelligent processing, a large amount of logical execution, can choose the Hadoop + hbase cloud computing solution.

MongoDB also supports JS map/reducer so you can try to integrate MongoDB into the cloud computing solution.

When I make a MySQL + nosql scenario, I will choose MongoDB, not only because of his excellent massive distributed solution support, nor because of the support of Map/reducer distributed computing. It's because I haven't heard of it. There have been major failure cases, relatively perfect documents (and Chinese manual yo) and JSON sub-lattice support, in the current Webapi popular era, not only from the personal love angle, but also from the project management point of view, the development of the more loved it, hehe.

I'll take the time to tidy up MongoDB related data sharing some, mainly to relive the previous efforts, convenient for their memories.

Nosql--mongodb selection