Scholarship & Research
Permanent URI for this communityhttps://scholarworks.montana.edu/handle/1/1
Browse
3 results
Search Results
Item Analysis and implementation: converging intent based production and high speed research networks(Montana State University - Bozeman, College of Engineering, 2019) Hess, Gregory Martin; Chairperson, Graduate Committee: Mike WittieThis thesis analyzes the performance impact of converging an intent-based production network with a high-performance research network. The previous solution for high performance research networks was to segregate these networks [1]. This solution created a physically separate network dedicated to the researcher and used for high speed data transmissions. This model has been successfully utilized for years however current refresh cycles are force academic institutions to confront the ongoing maintenance of these boutique networks. Some institutions have noted their investment in their production networks have created performance to rival that of the dedicated network. For these institutions convergence to one network proves to be a viable strategic option. Additionally, vendors are going to market with intent based or software defined networking which answers many of the challenges that required the physical separation of networks. The advantages of both converged networks as well as software defined networks are well documented. Both campus Information Technology departments and the researches with this high-performance needs are in need of a quantitative analysis to understand the performance or security trade-offs associated with moving research onto a production, intent-based network. This thesis addresses this question by measuring and comparing key performance metrics of a traditional high-performance research network, a traditional production network, and a converged Intent-Based network in the same three labs at the same institution (Montana State University). The results prove that a converged, intent based network delivers the same (or superior) performance as the previous model with the same or superior level of segregation (security). These results give institutions the ability to shed the traditional, utilitarian use of institutional networks in favor of a dynamic network model based on the identity and use of the network rather than the physical location.Item A local area network for intercomputer, interprocess data communications(Montana State University - Bozeman, College of Engineering, 1984) Schlegel, James LeroyItem Investigation of physically aware routing and wavelength assignment (RWA) algorithms for next generation transparent optical networks(Montana State University - Bozeman, College of Engineering, 2010) Hahn, Timothy Allen; Chairperson, Graduate Committee: Brendan MumeyOptical networks form the foundation of today's information infrastructure. Current generation optical networks consist largely of point-to-point electronically transmitted links which switch between nodes and repeaters. There is a trend in optical networking to move from the current generation opaque networks toward transparent networks. Transparent networks use only optical devices, eliminating the costly need for OEO conversions. Unfortunately, transparent networks present a unique challenge in maintaining acceptable signal quality levels. This research is an investigation of RWA algorithms in transparent optical networks. We present RAPTOR, a custom built discrete event program to simulate optical networks. RAPTOR uses its physically aware modules to accurately calculate three of the dominant physical impairments. RAPTOR is fast and multi-threaded. We introduce several new performance metrics. RAPTOR enables us to study transparent optical networks in a unique and realistic manner. We conduct an extensive performance analysis of existing RWA algorithms. We explore many different traffic models, traffic loads, signal quality, and network topologies in a comprehensive fashion. We directly compare the leading RWA algorithms in a manner has not been done before. We studied new RWA algorithms in two fields: Dynamic Programming and Ant Colony Optimization. Our new Dynamic Programming based algorithm has the best overall performance in most scenarios. It is flexible and adapts well to all network conditions we studied. It shows good promise for future optical networks.