ECLIPSE® Practice Management Software
Client Server Operations


Overview

Prior to version 8, ECLIPSE was only available in one flavor. All operations performed by ECLIPSE were performed directly. In other words, when ECLIPSE retrieved data from files, that data was transferred directly from the computer's hard drive to local memory used by the program. If that data resided across a network, large data volumes moving back and forth between computers could negatively impact system performance.

Along with various hardware and software improvements to increase network throughput, a new software paradigm took hold as a direct response to the fact that computer connections were much slower than the computers themselves.

The new model — dubbed Client/Server — presented a different strategy whose primary goal was to decrease network traffic. This new model called for two separate applications.

The Server application resides on the same network node as the database files. The sole purpose of this application is to fulfill requests to read data from and write data to the database. These requests are made by Client applications.

The client version of ECLIPSE sends requests for data and does not read data directly. For example, suppose you select a patient from the rolodex with 1000 account entries in her ledger. In v8 & v9, each read operation involves:

• Check the index to see if any (more) entries exist for this patient. This operation may require several disk reads.
   

• Read the actual information from the data file.
   

• Repeat step 1 while more entries exist.

The standard versions of ECLIPSE must therefore move a great deal of data across the network to retrieve each account entry. The Client version of ECLIPSE makes one request for this data by sending a message to the server. The server then returns all the information simultaneously. This potentially decreases network traffic by over 66% (the server checks the indexes locally and only passes back data) and gives a corresponding boost in performance. Naturally, this model has problems of its own (e.g. the overhead involved in sending messages). However, on a properly designed system, it should provide many advantages when under load.

Who needs it?

The best candidates for Client/Server are offices that have

• Huge data volumes (e.g. > 500,000 account entries on-line).
   

• 5 or more workstations in use simultaneously on a regular basis.
   

• Remote sites running the program.