What hardware is used to build large automated information systems?
Is it possible to design a system that provides acceptable performance under the condition of a large number of connected and likely concurrent users in it (48000 branches, each about 1-5 users connecting to the system + an indefinite number of client connections from the "outside", not a specific branch)?
In theory, this is a massive library of IP that implements functions: full-text database of literary works c search on it; interlibrary loan; electronic document delivery. Built on client server architecture.
It consists of data centers (which, in fact, spinning database) and is connected to the data center client computers (office located in the libraries and computers of readers who use the system from home). Work with the system via the web interface.
Is the bottleneck here, as far as I can tell, this full text database. First, you need to maintain a huge number of active user connections and process requests from them, secondly its purpose implies gigantic size of this database, further complicating the question of system performance.
In fact, the questions are the following:
1) What hardware, what hardware need to make this system work?
2) How can I organize the full-text database to provide performance at an acceptable level and to implement the functionality of storing high volumes of text search on them? To dig in the direction of distributed database, separating a few and distributing them on dedicated machines, for example?