RDBMS (Relational DBMS), which appeared in the 1980's, has continued to evolve in step with the changing IT environment. As a result, it firmly established itself as a core infrastructure of the enterprise information system. RDBMS, which uses disks as main data storage in most cases, has been extended its application area from basic database management to many other areas including data warehouse and data mining. In particular, the concept of separating data and application was very useful, and it led to the explosive growth of DBMS.
		
However, the DB application environment began to require high performance data processing due to changes in network devices and speed. In particular, the users whose number was quickly increased as a result of the rapid growth in the Internet and mobile communication wanted fast and stable real-time service regardless of time and location.

For instance, tens of thousands of users frequently connect to Internet portal sites or Internet game sites concurrently, and all of them want fast service. Especially, in case of the authentication system, the number of transactions the DBMS must process at the same time would be several hundreds. For another example in the mobile communication area, the subscriber location management system (HLR: Home Location Register) must process more than 15,000 transactions per second to handle 2,500 calls per second. Therefore, a DBMS providing fast processing speed and scalability, capable of handling a huge number of concurrent users, becomes essential.

In addition, along with fast response time and scalability for concurrent users, uninterrupted service provision is an important prerequisite for these application systems serving as the DBMS. Online stock trading systems, real-time billing systems and softswitch systems are typical database application services that must be provided for 24/7. They are typical fault-tolerant systems that should be interrupted in no case whatsoever. If the real-time billing system fails to compute charges even for a brief moment, the company will suffer a great deal of financial loss. Moreover, many real-time billing systems in mobile communication are designed to process 4,000 CDRs (Call Data Record) per second. Accordingly, these database application systems must guarantee fast response time of transactions while staying fault-tolerant.

The existing Disk-based RDBMS(or DRDBMS :Disk Resident DBMS), which has been favored by many for a long time due to its rich functions and convenient administrative tools, is now facing performance issues owing to the changes in the DB application environment. It is attributable to the fact that RDBMS uses disks as main data storage. Frequent disk I/O errors cannot but result in poor transaction processing performance.

To solve this problem, most RDBMS vendors attempted to tune and upgrade physical domain (I/O) and logical domain (buffer and memory). Even though the DRDBMS brings the data from the disk to the buffer of the memory to have faster processing speed, but it is impossible to store all of the data, which are frequently accessed by a lot of unspecified users, in the buffer. Accordingly, it is impossible to stay clear of the disk as the data storage space. Again, RDBMS is facing with the limitation of disk I/O.

Additionally, it was possible to ensure availability by means of database clustering for fault-tolerance. However, it was difficult to implement in areas requiring fast processing speed due to its poor performance from data negotiation and mediation. Accordingly, fast response time and scalability issues are yet to be resolved.

This led to the efforts to overcome the limitation of DBMS I/O. Especially, to overcome the limitation of the speed of disks, physical data storage, the discussions of DBMS utilizing memory, faster than disks, began. (Speed levels of storage media: Tape, Drum < Disk < memory < cache < CPU).
With focus on communication devices requiring high performance or areas sensitive to processing speed such as financial area, the necessity of real-time DBMS has surfaced. 
