Scalability and Data Transfer
We recommend that all application services be run in a private cloud environment. Many, if not most client users will connect to the private cloud using a wireless internet connection, so optimizing communications is an important factor in the scalability of the application:
- The amount of data transferred – this is important because it is a good indicator of the ability of the entire system to scale to many users. In large implementations like this, the network and the part of the system that sends the data over the network tend to become early bottlenecks in the overall performance of the system. Transferring less information per interaction will allow more to be done with the same resources. Additionally, since most wireless internet providers charge according to the amount of data transferred, the less data sent for each interaction, the lower the cost of maintaining the implementation will be.
- The time it takes to deliver the full screen to the client browser – this is tied to the “responsiveness” of the application in terms of time spent waiting by the users to see their search results, patient details, or to add new information.
In order to provide a comparison between Veracity and many commonly accessed web sites, we performed several tests and averaged their results:
|
The first test was a “time to load” comparison. For sites such as Google, Facebook and Twitter, we found their fastest loading pages (best case scenario) and used them for comparison. For Veracity pages, we took “difficult to handle” (searches that returned tens of thousands of records, very complete patient records, etc.) pages (worst case scenario) and used them. The times shown are in milliseconds. The VSI Veracity search returned 9,500 rows of patient data in the time shown. |
 |
| The second test shows the amount of data (in Kilobytes) sent from the server to the browser for the simplest of pages (theirs) and the most difficult pages (for Veracity): |
|
So, how does this relate to Scalability?Once the project has been installed and all users are trained, the data charges for the usage of the system will still be charged and either the country, the municipality or the local organization will need to pay for them. This is what has caused many projects in the past to fail – the equipment works well, the software meets local requirements, but the data charges are unaffordable. To provide a sense of relative costs, we tested the software we will use in these African implementations against several popular web sites and measured the cost of data that would be used to view each website 2 million times. We used a low-to-average internet data usage rate in our comparisons.
|
|
