Effective optical interconnect is a fundamental requisite to realize Internet-scale data centres due to the capabilities and benefits of optical devices. Optical interconnects are energy efficient and support massive bandwidths. The performance of optical interconnects is directly related to the type of optical switches and optical switching techniques used. Optical switches may be categorized into slow optical switches and fast optical switches. The optical switching techniques are Optical Circuit Switching (OCS), Optical Packet Switching (OPS) and Optical Burst Switching (OBS). This thesis presents three novel optical interconnection schemes which are based on optical burst switching. These schemes are called Hybrid Optical Switch Architecture (HOSA), HOSA with Traffic Demand Scheduling (TDS) and Fast Optical Switch Architecture (FOSA). The first two schemes are based on a hybrid design that utilizes fast and slow optical switches while the third scheme is based on using only fast optical switches. The proposed schemes consider OBS with a two-way reservation protocol that ensures zero burst loss. In the two-way reservation protocol, the connection is established for each burst before transmission. To evaluate the performance of these schemes, network-level simulation is used. The proposed architectures consider a single stage core topology that can be easily scaled up (in capacity) and scaled out (in the number of racks) without requiring major re-cabling and network reconfiguration. The proposed schemes feature separate data and control planes. The control plane comprises a centralized controller while the data plane contains an array of optical switches. A scalability analysis of the proposed topology is presented and shows that this topology is scalable to hundreds of thousands of servers. This thesis also presents a trade-off between cost and power consumption of the proposed designs by comparing them with conventional interconnects using analytical modelling. In HOSA, the key idea is to route high volume traffic through a fast optical switch during the reconfiguration of a slow optical switch. The traffic is moved to the slow optical switch once it is reconfigured. The aggregated traffic should be large enough so that it can bypass the slow optical switch during its reconfiguration phase. This technique hides the reconfiguration time of slow optical switch but the latency introduced due to burst aggregation is still high. In HOSA with TDS, small bursts are considered to reduce the latency of burst aggregation. The controller in HOSA with TDS technique maintains a traffic demand matrix which updates traffic demand on periodic intervals and assigns slow paths for the high traffic volume. A resource allocation algorithm is proposed that allocates paths of fast and slow optical switches efficiently. In HOSA with TDS, the control plane can only support applications that have high traffic stability. So for dynamically changing communication patterns, this thesis presents a new design called FOSA. The FOSA is based on only fast optical switches and it also uses OBS with the two-way reservation i.e. there is no additional burden on the control plane for maintaining traffic demands as was done in HOSA with TDS. The proposed technique shows considerable improvement in terms of throughput and packet loss ratio as compared to the traditional methods of OBS while performance comparable in terms of delay with the traditional methods of OBS is also achieved. The proposed technique also demonstrates performance comparable to that of electrical data centre networks. The performance of TCP over traditional OBS network is degraded by the bursts losses due to contention even at a low traffic load. In FOSA, the performance of the TCP is evaluated. Since in the proposed scheme, the burst loss is zero due to two way reservation, the results show significant improvement of TCP performance in terms of throughput, time and packets loss as compared to the traditional methods of OBS and the conventional electronic packet switching data centre networks for all types of workloads.