Many high performance computers (HPC) and cloud computing applications rely on distributing tasks among large numbers of virtual and real servers. This implies that advancements in performance of data centers and HPCs is increasingly dependent on connectivity. In order to insure high degree of connectivity at increasing bit rates and distances the demand for large bandwidth-distance product connections is increasing. These can almost exclusively be provided using optical interconnects. Traditionally optical-interconnect come in the form of pluggable transceivers. However the increases in number of connections and bit-rate poses a limit to further scaling (the front-plate bottleneck). A shift towards mid-board optics is in the making but requires solutions which are compact, power efficient and low cost for manufacturing. In this talk we will present our most recent demonstrations of high density optical interconnect solutions as well as high density switches. First some details about the design aspects and advantages of compact electronic switches employing mid-board optical engines will be discussed. Then, for addressing the challenge of low cost optical interconnects, we will give details on our recent work targeting high channel count VCSELs based sub-modules. Results based on 2.5D and 3D assembly on high resistivity silicon will be discussed as well as the use of direct die attach to flexible PCBs for making high density interconnects.
Traffic in data centers networks (DCNs) is steadily growing to support various applications and virtualization technologies. Multi-tenancy enabling efficient resource utilization is considered as a key requirement for the next generation DCs resulting from the growing demands for services and applications. Virtualization mechanisms and technologies can leverage statistical multiplexing and fast switch reconfiguration to further extend the DC efficiency and agility. We present a novel high performance flat DCN employing bufferless and distributed fast (sub-microsecond) optical switches with wavelength, space, and time switching operation. The fast optical switches can enhance the performance of the DCNs by providing large-capacity switching capability and efficiently sharing the data plane resources by exploiting statistical multiplexing. Benefiting from the Software-Defined Networking (SDN) control of the optical switches, virtual DCNs can be flexibly created and reconfigured by the DCN provider. Numerical and experimental investigations of the DCN based on the fast optical switches show the successful setup of virtual network slices for intra-data center interconnections. Experimental results to assess the DCN performance in terms of latency and packet loss show less than 10^-5 packet loss and 640ns end-to-end latency with 0.4 load and 16- packet size buffer. Numerical investigation on the performance of the systems when the port number of the optical switch is scaled to 32x32 system indicate that more than 1000 ToRs each with Terabit/s interface can be interconnected providing a Petabit/s capacity. The roadmap to photonic integration of large port optical switches will be also presented.
In this paper we report the design, fabrication, simulation and characterization of a novel discretely
tunable laser based on filtered feedback. This Integrated Filtered-Feedback Tunable Laser (IFF-TL) device
combines a simple and robust switching algorithm with good wavelength stability. It consists of a Fabry-Perot
laser with deeply-etched broadband DBR mirrors. Single mode operation is achieved by using feedback from an
integrated filter. This filter contains an AWG wavelength router and an SOA gate array. A rate equation model
predicts that a properly designed device can switch within 1 ns, while characterization measurements show a
value of only 4 ns. The fast switching and reduced control complexity makes the device very promising for various
advanced applications in optical telecommunication networks.
In this paper two different paradigms to realize a scalable
all-optical packet switch with label swapping will be reviewed.
All the functions required for switching the packets are based on
all-optical signal processing without any electronic
control. This allows very low latency and potential photonic integration of the systems. We report for both techniques
experimental results showing the routing operation of the 160 Gb/s packets and beyond. We will discuss and compare
both techniques in term of devices and bit-rate scalability, latency, power consumption, power penalty performance and
cascadability as key parameters for the realization of an all-optical packet switch.
We discuss how all-optical signal processing might play a role in future all-optical packet switched networks. We introduce a concept of optical packet switches that employ entirely all-optical signal processing technology. The optical packet switch is made out of three functional blocks: the optical header processing block, the optical memory block and the wavelength conversion block. The operation principle of the optical packet switch is explained. We show that these three functional blocks can be realized by using the nonlinearities of semiconductor optical amplifiers. Some technologies in these three functional blocks are described. The header processor is realized using a Terahertz Optical Asymmetric Demultiplexer. We also describe a header pre-processor to improve the extinction ratio of the header processor output. In the optical memory block, we show that an all-optical memory can be obtained by using two coupled lasers that form a master-slave configuration. The state of the optical memory is distinguished by the wavelength of the master laser. We extend the concept to an optical memory can have multiple states. In the wavelength conversion block, we demonstrate a 160 Gbit/s wavelength conversion using a single semiconductor optical amplifier in combination with a well-designed optical bandpass filter. The semiconductor optical amplifier has a gain recovery time
greater than 90 ps, which corresponds to a less than 20 GHz bandwidth for conventional wavelength conversion. We show that by properly using the optical bandpass filter, ultrafast dynamics in the semiconductor optical amplifier can be employed for wavelength conversion at ultrahigh bit-rates.
We discuss how all-optical signal processing might play a role in future all-optical packet switched networks. We describe a few approaches to optical header processing, all based on nonlinearities in a semiconductor optical amplifier. In first approach a SLALOM configuration is used. The second approach uses a Terahertz Optical Asymmetric Demultiplexer. We also describe a header pre-processor to improve the extinction ratio of the header processor output. The second functional block on which we focus is optical buffering. We show how all-optical signal processing technology can be used to route a packet into a fiber delay line and we describe a circulating optical loop based op optical technology.