Solutions
WAN Optimization Solutions
Certeon's WAN Optimization Solutions
Many IT managers try to mitigate WAN issues by just adding more bandwidth, only to discover that the effects of high latency and packet loss still impact remote application access performance, even during non-peak network times. This opened up the opportunity for multiple WAN optimization techniques such as compression, caching, and differencing, and products such as asymmetrical and symmetrical network devices to make their way into the market. Certeon's S-Series Application Acceleration Appliances approach to WAN optimization falls into the symmetrical class of devices, with some major differences in how Certeon optimizes the application traffic from other vendors.
Symmetrical WAN Optimization Techniques
The Certeon S-Series network optimization applications are deployed on each side of the WAN link to accelerate application response times. Each device incorporates a number of techniques to minimize the impact of WAN issues on application performance - all transparent to existing clients, servers, and applications. The S-Series WAN optimization techniques are discussed below.
Traditional WAN Optimization
Certeon's S-Series employs many traditional network optimization techniques, including the following:
- Packet Compression - Network packet compression works by reducing the packet size, thereby decreasing the need for WAN bandwidth. The S-Series uses a sophisticated proprietary history-based cache to difference packets against previous packets, achieving an average compression ratio in the 4:1 range.
- Traffic Prioritization - Using prioritization, traffic is separated into different categories, such as time-sensitive LOB applications, general business applications, background activities, and personal use.
- Protocol Proxy - A protocol proxy terminates connections locally on each side of the WAN and uses a WAN-optimized protocol and parameters for traffic control over the WAN. The S-Series HTTP proxy multiplexes requests over persistent TCP connections that are tuned to the actual WAN characteristics - using large window sizes and learned congestion control parameters. Persistent connections eliminate the long times required to set up new TCP connections over a WAN.
- Traffic Control and Forward Error Correction - Traffic control involves throttling the transmission of packets at the rate supported by the slowest network link - typically the remote office WAN link. Thus, low-priority packets can be queued, and dropped if necessary, at the source location, before clogging the WAN link, where high- and low-priority packets have an equal chance of being dropped.
Forward Error Correction (FEC) enables the receiving node to regenerate these lost packets without the extra latency of retransmitting them. With FEC, the sender adds one or more Error Correcting Code (ECC) packets to a group of data packets, and the receiver uses the ECC packets to regenerate any lost or significantly delayed data packets (up to the number of ECC packets). The S-Series utilizes both QoS and FEC for traffic control and error correction.
Application Intelligent Networking
The key challenge remaining is to significantly reduce the amount of Web data that needs to be sent over the WAN, thereby reducing response times and increasing the ability to support a greater number of remote users. Application-aware acceleration methods, such as Certeon's S-Series Application Acceleration Blueprints and Object Differencing Engine (ODE), provide much higher levels of data reduction and hence application acceleration. The Blueprints identify specific application objects, and the ODE maximizes the differencing between those objects and sends only those differences over the network. No changes are made to the client or server software or system - acceleration is fully transparent.
Key attributes of Application Intelligent Networking include:
- Fast differencing based on application awareness - With application awareness, the stored history of each data object used by applications can be separately aggregated and versioned so that differencing need only occur against what is known to be highly correlated content.
- Larger data-object sizes, enabling better data reduction - Most application data objects, such as files, lists of database records, and Web objects, are tens of kilobytes to tens of megabytes in size compared to packet sizes of 0.5 to 1.5 kilobytes. Larger buffers allow much faster optimized difference and compression algorithms.
- Large history store - The total history store is minimally 100 GB to maintain a week's worth of T1 data so that the data can be repeatedly referenced to reduce WAN traffic. This translates into a million objects averaging 100 KB each. The smaller caches typical of packet-level-only compression are quickly overloaded and unproductive except in simple repeated download tests.
- Predictive preloading - Application awareness enables the predictive preloading of historical content that will subsequently be used for differencing, without adding disk access latencies to the overall response times.
- Intelligent history management - Application awareness allows related data objects to be grouped together for fast access, flexible aging, and fast object content learning and synchronization.
Certeon's S-Series enables 1,000 percent improvement in application response time through its unique network optimization techniques. Learn more about Certeon's outstanding performance in WAN optimization and application acceleration.
