Features

High scalability

Using cluster construction of multiple computers to create replication and share load, Kokolink PostgreSpread provides a way to protect data and at the same time, a high performance search functionality according to the number of computers using.

• Replication node
  To improve the reliability, the same data is saved to two different hosts. Then, database service is still available even when one host has trouble and becomes failure.
  

• Mirror node
  To provide high performance with the use of load sharing, multiple mirrors share the same data for read, write or query.
  

Implementation of PostgreSQL property

Kokolink PostgreSpread has the high compatibility of PostgreSQL 8.1, allowed to create systems with all PostgreSQL rich and robust applications and tools.

Low cost system construction

The system can be constructed of three server machines with intranet network. Sharing hard disks or expensive hardware such as SAN are not required.

Reliable integrity protection

Kokolink PostgreSpread uses two servers, master/slave, with a number of mirrors as replication. By using random numbers, time, etc..., it allows to create accurate copies of data from server original data to database information such as OID, CTID, etc...

Database copies are made every transaction commitment. Transaction data that committed will be safe even when one server machine has trouble.

Mirror data references are synchronized as well. The same data will be available on mirrors just after it has committed to master host.

High availability

Master and slave observe each other at heart-beat pace. If failure occurs, failover will be executed automatically. Down time due to failure is then significantly minimized. Kokolink PostgreSpread also has the IP switch function. If failure happens, IP address switching will be operated. And the same IP address can be used to access master database after failure.

Mirrors also switch to the new master upon the failover of master/slave. If failure happens on some mirror, that mirror only will be off, and the whole system keeps on running normally.

By taking advantage of these functionalities, client can access database without any consciousness of system failure.

Easy system operation

Cluster system is constructed automatically when server starts. Cluster is built-in function of Kokolink PostgreSpread, therefore there is no need of other middleware. When a server started, it reach for other server on the same network, and decides its own role automatically. After that, database synchronization takes place and cluster system starts.

Shutting down or adding individual server can be taken place without suspending the whole system. Therefore, there is no need for suspension due to updating servers or switching between master and slave.