HBase Anti-Patterns
What tasks cannot be used for HBase?
I understand that HBase and HDFS should be treated as temporary storage stores, storing data only for the time that the map / minification is required.
Can't HBase be used as a canonical data store? The frequency of random access latency makes this pretty impractical anyway, but this can be mitigated with caching and other useful features.
source to share
HBase is supposed to be used as a canonical data store (for example, Big Table, which is used in many Google services). HDFS is built for MapReduce, but HBase is built on HDFS, which allows more than just MapReduce. HBase is truly a database.
The main reason you should choose HBase for traditional relational database systems is scalability. If you don't have big data, don't worry about using HBase. If your application needs a lot of connections, it needs a query language like SQL and it doesn't scale data around the world, prefers RDBMS.
On the other hand, if your relational database has a huge number of NULL records, HBase is a good alternative since it is sparse: it does not store NULLs.
source to share