In this article we present a review of the latest activities in recent experimental high-performance optical networks such as ultrascience network (USN), dynamic resource allocation via GMPLS optical network (DRAGON), and circuit-switched high-speed end-to-end transport architecture (CHEETAH). We compare the control and management approaches adopted in each of these networks and analyze their capabilities vis-a-vis the functional requirements of grid computing applications. Grid computing is increasingly on the rise to meet the massive processing and storage demands of a new class of e-science physics applications that may generate and require the processing of data sets reaching terabytes per day. The requirements of these applications challenge the limitations of the networking technologies that are in place today. In particular, the area of network management and control is undergoing significant developments in order to meet the demands of these applications. It is the purpose of this article to share our experiences in the deployment of the GMPLS control plane in these experimental optical networks. It is our belief that these and similar efforts will result in significant progress toward enabling connection-oriented high-performance networking. This new paradigm will encompass grid computing applications as well as commercial, health, and entertainment services, thus making it useful to the public at large.