MYSQL,HBASE,MONGODB implementation for data sequential storage

The Indexed organization table for mysq,oracle is the data stored in the primary key order (in-page).

If the data insertion order is inconsistent with the primary key, there is no guarantee that the pages will be stored in a physical order. The cost of moving data blocks is significant. The storage structure can be re-sequenced by means of tools.

When MySQL adds data blocks, it will leave a certain amount of space for future new data, and no additional data will be allowed when the data block fills a certain percentage.

Because of the need to set aside some space for the update data use (mainly to adapt to the varchar field changes). are stored sequentially in the page. So MySQL data block size, the idle scale setting directly affects the extent of page fragmentation.

HBase and MongoDB use sstable,lsm tree-related techniques to regularly organize and merge data blocks, so data blocks are stored in primary key order. There will be no fragments. The system impact is also minimal during the merge.

HBase storage When the data is compressed or even compression, occupies little resources, saving a lot of Io, memory. There are a lot of gaps in the storage scenarios for MySQL and Oracle.

HBase has an advantage in insertion and sequential scanning efficiency.

In addition, HBase's column storage efficiency should be higher than the MongoDB document way. However, since mongodb,hbase is compressed for sparse data. So it's not as big as you think.

MongoDB is similar to the relational library because it supports level two indexes. There are more scenarios to apply.

MYSQL,HBASE,MONGODB implementation for data sequential storage