The BaBar data acquisition system[HSC$^$99] is layed out similar to the proposed CMS architecture (see section 2.3.2). A number of VME based processors read out and buffer data that come from the detector electronics. These data are shipped to a processing farm by using a non blocking switching network. The efficiency requirements of this system are however lower than for our application. BaBar requires its data acquisition to digest events at a size of 25 KBytes with 1.3 kHz. This is roughly two orders of magnitude smaller than the planned Large Hadron Collider experiment described in this work. In addition to event filtering the on-line software in the computing farm is also responsible for performing data compression. The BaBar experiment has adopted the object-oriented methodology for the on-line software subsystem. This does not include the use of object-orientation at system programming level, but at the boundary to the application programmer. A framework[HHM$^$98] that shields the physics programmer from all low level details is provided for using the services of the readout module and the processing farm. The developers of this system however found it difficult to realise all necessary performance optimisations[SCG$^$98] without sacrificing the abstraction obtained from applying the object paradigm.