Rapid growth in eBusiness has made industry and commerce increasingly dependent on the hardware and software infrastructure that enables high-volume transaction processing across the Internet. Large transaction volumes at major industrial-firm data centers rely on robust transaction protocols and adequately provisioned hardware capacity to ensure that the high-volume flow of daily business transactions is not interrupted. This research analyzes the robustness of eBusiness transaction infrastructure using Petri nets and simulation to characterize two critical elements that were found to be problematic in industry: 1) a widely used electronic messaging protocol for eBusiness transactions called RosettaNet (RN), and 2) capacity planning for the bottlenecked application-server tier of hardware infrastructure used to process eBusiness transactions. Analysis of the RN protocol is based on reusable patterns for Petri net modeling of common protocol mechanisms such as retries, time-outs, and fault handling. These new patterns translate abstract representations of the RN standard into an executable model. A stochastic Petri net constructed from the patterns is simulated to generate performance curves based on varying network conditions, and analyzed using linear temporal logic to prove that the protocol can end processing in unsynchronized failure states that make resolution of the interchange problematic. This leads to suggested improvements in the protocol. Analysis of the application-server tier of the infrastructure used for processing eBusiness transactions is also performed to enable better capacity planning for robust operations. Improved planning for this infrastructure requires characterization of resource consumption patterns driven by two key factors affecting service performance and capacity: 1) the varying transaction load patterns, and 2) the variable delays incurred by downstream application services. Both factors are considered because servers consume and hold computing resources both when transactions arrive and also while requesting and waiting for responses from external services needed to fulfill transaction requests. A two-step, emulation-simulation approach is used that empirically determines consumption patterns for a single test server by emulation testing under controlled load and latency conditions; then data from the emulation tunes a scaled-up discrete event simulation model to predict production server consumption. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://www.proquest.com/en-US/products/dissertations/individuals.shtml.]