Invented by Deepak Goel, Pradeep Sindhu, Srihari Raju Vegesna, Robert William Bowdidge, Ayaskant Pani, Microsoft Technology Licensing LLC

The market for data center networks with multiplexed communications of data packets between servers is experiencing significant growth and innovation. As the demand for data storage and processing continues to rise, data centers are becoming increasingly crucial for businesses across various industries. To meet this demand, data center networks are evolving to provide faster, more efficient, and reliable communication between servers. Multiplexing is a technique that allows multiple data streams to be transmitted simultaneously over a single communication channel. In the context of data centers, multiplexed communications enable the efficient transfer of data packets between servers, optimizing network performance and reducing latency. This technology is particularly valuable in large-scale data centers where thousands of servers are interconnected. One of the key drivers of the market for data center networks with multiplexed communications is the exponential growth of data generated by businesses and consumers. With the rise of cloud computing, artificial intelligence, and the Internet of Things (IoT), data centers are under immense pressure to handle massive amounts of data in real-time. Multiplexing allows data centers to handle this data influx efficiently, ensuring smooth operations and minimizing bottlenecks. Another factor contributing to the market growth is the increasing adoption of virtualization technologies in data centers. Virtualization enables the consolidation of multiple servers into a single physical machine, reducing hardware costs and improving resource utilization. Multiplexed communications enhance the performance of virtualized environments by efficiently managing data traffic between virtual machines. Furthermore, the market for data center networks with multiplexed communications is driven by the need for high-speed and low-latency connections. In industries such as finance, telecommunications, and healthcare, where real-time data processing is critical, multiplexing ensures that data packets are transmitted quickly and reliably. This is achieved through advanced multiplexing techniques like wavelength division multiplexing (WDM) and time division multiplexing (TDM), which allocate bandwidth efficiently and prioritize time-sensitive data. To meet the growing demand for data center networks with multiplexed communications, technology companies are investing heavily in research and development. They are developing innovative solutions that combine multiplexing with other cutting-edge technologies such as software-defined networking (SDN) and network function virtualization (NFV). These advancements enable data centers to dynamically allocate network resources, optimize traffic flow, and provide greater flexibility in managing data packets. In conclusion, the market for data center networks with multiplexed communications of data packets between servers is witnessing significant growth due to the increasing demand for efficient data storage and processing. Multiplexing technology plays a crucial role in optimizing network performance, reducing latency, and improving resource utilization in data centers. As businesses continue to generate and rely on vast amounts of data, the need for high-speed and reliable communication between servers will only intensify. Therefore, companies investing in the development of multiplexing solutions are well-positioned to capitalize on this growing market and cater to the evolving needs of data centers.

The Microsoft Technology Licensing LLC invention works as follows

A network system is described for a datacenter in which a fabric switch provides interconnectivity so that any server can communicate packet data with any other server using any number of parallel paths. According to the techniques described in this document, edge-positioned switch fabric access nodes can be configured in such a way that they provide full mesh interconnections, with only a single L2/L3 hop between any pairwise combinations of access nodes. This is true even for massive data centers containing tens or thousands of servers. Access nodes can be arranged in groups of access nodes, and permutation device may be used to spray packets over the groups before injection into the switch fabric. This increases the fanout and scaleability of the system.

Background for Data center network with multiplexed communications of data packets between servers

In a cloud-based datacenter, a large group of servers interconnected provides computing or storage capacity to execute various applications. A data center, for example, may be a facility which hosts services and applications for customers, or subscribers. The data center can, for instance, house all infrastructure equipment such as compute nodes and networking and storage systems. It could also host power systems, environmental control systems and other systems.

In most data centers clusters of storage and application servers are connected via a high speed switch fabric, which is provided by one or several tiers physical network switches and routing devices. The size of data centers can vary widely. Some public data centers contain hundreds of thousands servers. They are typically distributed over multiple geographical locations for redundancy. A typical data centre switch fabric includes several tiers of interconnected routers and switches. “In current implementations packets between a server source and a server destination or storage system will always be forwarded along a single route through the switches and routers comprising the switching fabrics.

Several data center network configurations and architectures are described. According to some examples of the techniques described, access nodes coupled with the servers, intermediate (e.g. electrical and/or optic) permutation devices, and core switch of the switch fabrics may be configured and organized in such a manner that the parallel data pathways in the switch material provide full mesh interconnections (any-to any) between any pairwise combinations of the access nos, even in huge data centers having hundreds and thousands of servers. The permutation device couples the access switches to the core switches via optical links in order to transmit data packets as optical signals between the access switches and core switches. Each permutation device has a set input ports and output ports that are used to send signals between the core switches and the access nodes to communicate data packets. The permutation device is configured so that the communications from the input ports are permuted between the output ports according to wavelength. This allows full mesh connectivity without optical interference between the ports facing the edge and the ports facing the core. In some cases, the permutation devices can be replaced with an electronic switch or router. The permutation device can be used with electronic switches or routers in other implementations.

This disclosure is directed at an optical permutor in some examples that acts as an interconnect to transport optical communications between devices on a network, including devices inside a datacenter. The optical permutor described herein includes a plurality input ports for receiving respective optical input signals. Each optical input signal may carry optical communications in a variety of wavelengths. The optical permutor’s internal optical elements are configured so that optical communications from input ports can be “permutated”. The optical output ports are mapped based on wavelength to ensure full mesh connectivity and to prevent optical interference caused by wavelength collision. The optical permutor ensures that communications from any optical input port can be sent to any optical output port without interference. The optical permutor can also be bi-directional. The optical permutors can provide full mesh bi-directional connectivity between points for optical communications.

The system can also include a number of servers, a number of access points, each access point coupled to a specific subset to the servers, to communicate data packets; an electrical permutation unit coupled to the subset and configured to communicate data to the other access points within the plurality, where the electrical device has a set input ports and output ports for communicating data between the subset, where each input port receives packet flows of data that have unique source addresses, and the electrical device is configured permute the plurality packet flows of packet flows of packet flows of packet flows of packet flows, each a

The disclosure also describes another method that involves interconnecting multiple servers through an intermediate network that includes: a number of access points, each access point coupled to one subset within the plurality, and an electrical permutation unit that is coupled to the subset and configured to communicate packet flows to other access points, wherein, the electrical device has a number of input and output ports that are configured to communicate packet flows between subsets of access points, and each input port receives packet flows of a variety of packet flows, with each packet flow, each having a

The drawings and description below provide details on one or more examples. The description, drawings and claims will reveal other features, objects and advantages of the invention.

BRIEF DESCRIPTION DES DRAWINGS

FIG. “FIG.

FIG. “FIG.

FIG. “FIG.

FIG. “FIG. 3.

FIG. “FIG.

FIG. “FIG. 4.

FIG. “FIG.

FIG. “FIG.

FIG. “FIG.

FIG. “FIG. 1.

FIG. “FIG.

FIG. “FIG.

Click here to view the patent on Google Patents.