Data centers are large facilities that house tens of thousands of interconnected computing and storage servers. In recent years, large investments have been made in massive data centers to support services such as search, social networking, electronic commerce, and cloud computing. This has spurred a lot of interest in the industry and the research community in innovation for reducing costs and improving the performance of data center networks. One of the most crucial components of a data center network (indeed, of any network) is its transport mechanism -- the method by which data is transferred from one server to another. The major goals of a transport mechanism are to transfer data at the highest possible rate and with the lowest latency. In this dissertation, we present measurements from three production clusters with 6000 servers that reveal impairments with today's state-of-the-art Transmission Control Protocol (TCP) in data center environments. The impairments are rooted in TCP's demands on the limited buffer space available in commodity data center switches. We then propose two novel transport mechanisms designed specifically for the data center: DCTCP and HULL. These mechanisms enable the construction of large-scale, high-performance data center networks with very low latency and high bandwidth using commodity switches. DCTCP (for Data Center TCP) uses a simple modification to the TCP congestion control algorithm that allows it to maintain very low queue occupancy (and latency) in data center switches while simultaneously providing high throughput. HULL (for High-bandwidth Ultra-Low Latency) is an architecture that builds on DCTCP to deliver baseline fabric latency (only propagation and switching) by nearly eliminating all queueing from the data center network. We also present a stability analysis of the Quantized Congestion Notification (QCN) algorithm that has been standardized as the IEEE802.1Qau standard for Layer 2 (Ethernet) congestion control.