Latency Reduction for eGaming Website

Latency is always an issue for any live dealer gaming platform or live in play betting operator sites.

Zero latency is what is actually needed but every milli or even microsecond of incremental latency is the difference between making profits or not.

If your eGaming operator business depends upon reducing latency to achieve more profitability then we have the answer for you. Computrad work with leading technology providers utilizing technologies from industry leading partners. Computrad can reduce latency to accelerate the movement of data to create competitive advantages for your business.

How is this achieved in the data centre ? Utilising the following technologies:-

Wirespeed Throughput

Utilising 10GB Ethernet technology can allow the forward on all ports at Layer-2 and Layer-3 at wirespeed while simultaneously switching and routing unicast and multicast traffic. There are eight linecard modules in the system and each linecard provides 720 Mpps of throughput. Aggregated together, the system can forward 5.76Bpps. The linecard modules are always non-blocking. The ports forward at full line rate in all scenarios, at all packet sizes - small and large.
Port-pairs: With up to 192 pairs, all ports can send to each other without affecting performance.

Full-Mesh: Performance is not impacted based on full-mesh traffic patterns in the high end models. Even with a full mesh, any-to-any workload it still forwards without dropping any frames.

Low Latency

The low-latency virtual output queue fabric (VOQ) and an advanced traffic scheduler results in minimal delay in forwarding. All packets are received and stored on the ingress packet processor, which is equipped with very deep buffers. In parallel, while storing the packets, a forwarding decision is made by the scheduler as quickly as possible and the packet is divided and sent across the fabric to the egress port(s). The packet is re-assembled and transmitted out in most cases without any delays. This results in a very low latency - under 5 microseconds on average, for standard Ethernet frame sizes. In some cases nanosecond latency can also be achieved.

10 Terabit Lossless Fabric / Deep Buffer Congestion Management

With 384 MB of packet buffer for every eight ports, data gets queued when the egress interfaces are experiencing wide fan-in of traffic, or when supporting 10Gb to 1Gb transitions. The deep buffers ensure that congestion is absorbed and the egress interfaces are always working at their most efficient. This minimizes retransmission and results in much better performance at the application level. This ability to handle congestion, along with hardware support for Priority Flow Control (PFC), makes the system well suited for all applications that require a lossless fabric such as IP Storage, NAS, and video origination services.
One of the switches Computrad offer comes with 6 fabric modules. Each fabric provides 1.66 Tbps of switching capacity. Only 5 fabrics are needed at any given time to provide full non-blocking performance.
Another series of 10 Gigabit Ethernet switches offer up to 48 ports of non-blocking bandwidth at line rate, high-performance, low-latency switching with the highest port density in a 1 RU form factor.



Copyright © Computrad (Europe) Holdings Limited - All Rights Reserved - Our Privacy Policy