“
For many years, in order to improve the database server Performance, database administrators have to purchase larger servers when the database load increases (scaling), instead of distributing the database among multiple “hosts” as the load increases (scaling). RDBMS usually does not scale easily, but Newer NoSQL databases are actually designed to scale easily to take advantage of new nodes, and are usually designed as low-cost commodity hardware. “
I am confused about the scalability of RDBMS and NoSQL.
< p>My confusion is:
Why is RDBMS unlikely to expand? The reason to buy a larger server instead of a cheaper one.
Why can NoSQL scale?
NoSQL solutions usually provide record-level atomicity, but there is no guarantee that a series of operations will succeed (transactions). /p>
It boils down to: In order to maintain data integrity and support transactions, a multi-server RDBMS needs to have a fast back-end communication channel to synchronize all possible transactions and writes, while preventing/handling deadlocks.
This is why you usually only see 1 master (writer) and multiple slaves (readers).
I discussed the advantages and advantages of NoSQL in the technical blog Disadvantages, I have read the following
“
For many years, in order to improve the performance of the database server, the database administrator had to buy a larger server when the database load increased (expanded), and The database is not distributed among multiple “hosts” as the load increases (expands). RDBMS usually does not scale easily, but newer NoSQL databases are actually designed to scale easily to take advantage of new nodes, and are usually designed to be low Cost commodity hardware. “
I am confused about the scalability of RDBMS and NoSQL.
My confusion is:
Why is RDBMS unlikely to expand? The reason to buy a larger server instead of a cheaper one.
Why can NoSQL scale?
RDBMS has ACID (http://en.wikipedia.org/wiki/ACID) and supports transactions. Due to these concepts, the use of RDBMS for extension becomes More difficult to achieve.
NoSQL solutions usually provide record-level atomicity, but they cannot guarantee that a series of operations will succeed (transactions).
It boils down to: In order to maintain data For integrity and transaction support, a multi-server RDBMS needs to have a fast back-end communication channel to synchronize all possible transactions and writes, while preventing/handling deadlocks.
This is why you usually only see 1 A master (writer) and multiple slaves (readers).