Throughput Control of Message Based Integration Systems with Microsoft BizTalk Solutions

OData support
Huszerl Gábor Tamás
Department of Measurement and Information Systems

In my thesis I would like to present the basis of the modern enterprise service oriented infrastructure, based on Microsoft software, as well as I pay attention for traffic control, a distinguished problem in this sector.

Nowadays, a system can be so large that it is often necessary to split parts of the software. The various features and modules that are capable of running independently should be developed separately and be operated independently. Thereby, this creates a much clearer system that is separated in smaller parts which is flexible, fast and by the appropriate settings, also safe.

These services need to be gathered together and a work procedure is needed to be made, that is capable of solving the original problem by calling one after the other. We can select from many systems to include different services and to perform a work procedure that calls for different services in a particular order with different parameters. I choose BizTalk, a Microsoft product, which is very common and well documented.

BizTalk receives and forwards messages towards different services. It is possible that during the transmission of the message one of the services is overloaded and it is not able to receive as many messages as BizTalk would like to send; in this case the slow end point is completely flooded with messages. This might lead to failure of the end point or drastic slowdown. I worked out a solution for this problem, which I would like to present in my thesis.

I examined different traffic control methods in practice, after that I developed a well-functioning solution for this problem. My thesis is mostly based on the leaky bucket algorithm. Based on this, I created a traffic controlling algorithm and work pro-cedure that works with BizTalk.

Several different implementations occurred; however, in this thesis two of them were. Both of them use SQL server and works by the help of two orchestrations.


Please sign in to download the files of this thesis.