04/23/2001

>

> 1. In order to answer #1, wouldn't we have to address #2? Or do you

> just want us to mention which type of client-server that should be used?

> Also, if I chose to use the two tiered client server (the basic client

> and server), that means that all the mini databases from the 18 location

> is combined into one whole database that sits on the server, is that

> right? Not 18 separate databases sitting on the server?

I defined client-server class as the situation where functionality is distributed but data remains centralized. For question #1, you should assume that the database will be centralized and is located at the branch in Oakland. I'm only asking you how you would distribute functionality between the two types of clients and the server(s).

>

> I guess my problem is that I feel as though the two specifications that

> are given for assignment 9 is conflicting. If the return pc can check

> back in any book from any location, that means all the information for

> the 18 location should be combined into one large database.

This is incorrect. It simply means that the return PC has to have access to the database when a book is returned that belongs in a different location. It doesn't restrict you to having only one database.

> But if that

> is the case, how would a patron search for a book by one location? I've

> read on distributing database, thinking that's where the answer lies,

> but I'm still not grasping this problem. If there is the combined

> database on the server and the replicated minidatabase for each location

> sitting in the local machines, how would a patron search a book in not a

> current location but in another location?

>

Again, it's a question of access. When you distribute a database by fragmentation, it doesn't mean that each location only has access to its fragment. All locations should still be able to access the entire database. It's just that accessing a piece of information that has been stored at a different location requires (1) knowledge of what location has the information you need and (2) the network has to be able to handle both the request and sending the resulting information back.

> Could you give me a clue as to how I'm supposed to look at this? I'm

> worried that I'm totally off.

Your goal with question #2 is to determine how best to distribute the data across the entire library system based on the two major areas provided. You should assume that all locations will have access to all data no matter how it is replicated or fragmented. Your goal is come up with a way to distribute the data that minimizes the amount of network traffic and other overhead associated with processing distributed data.