Behavioral Traffic Control
By maintaining awareness and insight on all flows passing through it, the FR-1000 adeptly manages the diverse mix of traffic in a way that best matches the network's established policies for service delivery and quality. Unfortunately, IP networks have been hard pressed to deliver anything more than the most rudimentary forms of Quality of Service (QoS) given the total lack of flow state insight. Packet networks without flow insight are ill equipped to successfully manage today’s rich media network traffic. This has led to the market growth of specialized WAN Acceleration and Optimization appliances designed to offset the shortcomings of traditional packet networks by performing any number of optimization/acceleration functions based on flows, situated "upstream' of the router on the LAN as a separately running process. These products may perform some or all of the following functions aimed at improving end-to-end application performance:
- Traffic shaping (altering delay between traffic types by shuffling packets by class)
- Compression - replacing recurring data patterns with symbols to increase throughput
- Caching - placing popular content throughout the network and closer to destinations
- Data reduction - only sending "changed" traffic over the network while locally caching static unchanged data
- TCP "spoofing" and window size manipulation - locally sending expected TCP packets that would normally travel all the way between host and client, maximizing window size to eliminate overhead and delay
- Other application-specific mechanisms to accelerate application performance over the WAN
The presence of these products underscores the severe deficiencies in current-day networks. Without flow state information, traditional networks lack the real-time insight into how to optimally manage quad-play traffic. Anagran's Behavioral Traffic Control™ (BTC) changes that by optimally managing all traffic by flow class at the most efficient, cost-effective points in the network – wherever congestion occurs.
BTC basically '"watches" every flow. By constantly comparing each flow's behavior over time against a simple set of operator-defined rules per flow class, BTC can identify "suspect" flows that by virtue of their duration, byte count, source/destination, or other criteria, require some form of corrective or policing action. Thus, BTC can take any of the following actions, including
- reduce the allowed maximum rate of the flow
- demote the flow to a lower class
- promote the flow to a higher class
- direct the flow over another output port
Given the enormous growth of P2P and other forms of high-volume user-to-user traffic primarily in the form of video sharing, the value of BTC becomes clear. For the first time, the network has the ability to optimally manage all traffic in real time by class, regardless of the traffic mix. Since the FR-1000 is a wire-speed 96 Gbps device, this kind of traffic control by class is achieved at unparalleled levels of throughput and consistently low latency for all traffic. And unlike common traffic control mechanisms, BTC does not halt or eliminate any traffic types, it merely manages the entire traffic mix to allow all forms of network traffic to “peacefully co-exist”.
