Host Pacing technology

SunGate removes all congestion, drops and retransmission from the network. It does this by effectively pacing traffic of all types to the link speed and unlike the competitors it does this without the need to queue traffic.

SunGate’s host pacing technology emulates seperate links per user , controlling throughput bidirectionally to the bit and window shaping / ack pacing across these to further control congestion. Most competitors do not do use TCP window anipulation at all! This is the only natural way to reduce the number of packets on the network at any given time and therefore reduce congestion allowing higher prioirity traffic a free passage through the network.

The fastest way for an IP packet to traverse any link is when there is zero congestion between the client and host. I.e.: no packet queues in the router or anywhere along the route from client to server. Applications then travel at wire speed.

By entering the WAN link speed into the SunGate device its fair allocation of bandwidth algorithms will automatically and fairly split bandwidth amongst the users. This is often all that is required to greatly increase the performance of the network and improve the end user experience.

One can bias performance and create tiered services per user by using the rulesets, dynamically pacing non latency sensitive protocols further and allowing high priority traffic to always have a free lane through the network.

Almost all Internet communications have a client/server model where the client is sending requests and the server is sending data. This is true for ftp transfers, streaming video and streaming audio. Even if the client and server are sending UDP packets there is always a client/server relationship of some form.

SunGate paces hosts by controlling client requests so that higher priority traffic always has a free path through the network and never gets caught in a queue or buffer shuffle. It is proactive in that it paces all traffic to the speed of the WAN link so that there is no congestion at all along the traffic path. It monitors the flows of traffic and dynamically ensures high priority traffic has the bandwidth it requires.

This is radically different from the queuing methods employed by Cisco’s p-cube , Allot’s per flow queuing – and other so called “packet shaping tools”. It is the only method that allows you control over Internet traffic coming into your network.

The biggest advantage to pacing client requests is that you get at the source of traffic so that traffic flows at the given wire speed without forming queues or buffers. This results in much happer clients and much higher contention ratios are achievable without degragation to quality.

Queuing methods used by competitors are reactionary. Once the link “fills up” and reaches the given wan speed , then packets can no longer traverse the WAN at link speed and therefore enter the routers / shapers buffers and are Queued in order to be prioritised. Only Outbound traffic is queued. By queuing you are adding latency as you have to delay the packet in a buffer in order to prioritise it and let it back onto the wire. Additionally queuing methods can do nothing about the rate of incoming traffic. Within an ISP this is the majority of traffic.

Competitors will argue, and it is true that TCP and UDP will react to this congestion (given time) themselves but the point is that there is already congestion which is principally what you are trying to avoid if you are concerened about the traffic quality. Buffers overflow and traffic is dropped on the floor and in the case of UDP packets are lost forever or retransmitted adding to congestion if its a TCP stream.