Remote SQL Mode & Local Memory Mode
Alinous Elastic DB has Remote and Local Mode. In remote mode, it works as distributed RDBMS. In local mode, it works as high speed database.
This database engine works on only remote server. Then it works as high speed database which works mainly on memory.
The version written in C++ has own memory management system, and Garbage Collection engine. They cooperate together to handle data first.
This mode is for usage which needs high speed response time.
The database has strong buffer between the transaction processing engine and file I/O interface. If the server has enough memory, it works on memory.
But if there is not enough memory on the server, the cache engine uses LRU(Least Recently Used) algorithm. Therefore algorithm like BTree can work with less cache miss hit.
The memory management system handles memory allocation. Each thread has own memory pool to allocate. Therefore, memory allocation is very fast.
Huge number of row lock needs big memory. Then this database engine swap the locks from memory to disk.
Therefore lock escalation does not occurs.
The programming interface & all user interface is compatible with the database used as distributed database. Therefore, by only changing cloud node structure, or mode, it becomes scalable.
High Speed Database As Storage Engine
When this database engine works in distributed mode, it has storage nodes. The storage engine reuses this mode's components.