|
Web applications have become richer over time, including more interactive data with larger embedded images and file attachments. In addition, many client/server applications have a substantial amount of code that must be downloaded to the client browser before operation. All of these attributes make it increasingly difficult to deliver web applications across a WAN.
The performance of web applications is further debilitated by programming inefficiencies. Up to 80% of web content generated by enterprise applications is treated by web servers as dynamic content, though the actual underlying content is often static. Dynamic content is not cacheable by the browser or intermediate cache proxies, it is re-generated by the application server and repeatedly sent across the WAN to client browsers every time it is requested. This hinders application performance. (Click here to learn more about why web caches sometimes have difficulty caching).
Silver Peak appliances accelerate and secure all Web-based enterprise applications and portals through several optimization techniques:
WAN Deduplication
Silver Peak appliances capture all ‘information streams’ that traverse the link and recognizes that the information is identical. The first time that an HTTP object (static or dynamic) is requested, the information is compressed and sent over the WAN. When a browser later requests the same object from a server that dynamically creates the content, a Silver Peak NX appliance recognizes the ‘information stream’ and delivers the information locally. Even redundant traffic from dynamic content is always eliminated from the WAN. In addition, if a file is sent via email and later downloaded using HTTP, the Silver Peak NX Series appliance will detect the identical content and deliver the content locally.
Network Memory remembers and recalls byte stream patterns, not ‘layer 7 objects’, which is an important distinction from web caches. In addition, by operating at the network layer and remaining application transparent, Network Memory™ benefits all applications, not just TCP applications and protocols like HTTP. With Silver Peak, every client request is passed through to the server and the data received by the client is always fresh, not a cached copy that can be out-of-date. As a result, Silver Peak always ensures 100% data coherency.
Improving Latency
HTTP 1.0 required a TCP session to be opened, Acknowledged and closed for every object being requested by a browser. A page with 50 embedded GIF images, for example, required 50 separate TCP sessions to be opened and closed, sequentially. HTTP 1.1 overcomes this chattiness by implementing ‘keep-alives’ and pipelining (parallelization), which together dramatically reduces TCP protocol traffic. Silver Peak uses additional Network Acceleration tools to improve HTTP performance, including window scaling, selective ACK, and fast re-transmit. This speeds up web portals and business critical web applications, particularly when they are accessed over higher latency links.
Packet Coalescing
Silver Peak uses packet coalescing to re-package multiple smaller packets into a single larger packet. This is particularly beneficial for Web applications, like Oracle, which use J-Initiator – an Oracle java runtime that runs in the browser to provide a more robust client. J-initiator creates a stream of small packets on the WAN, each of which requires a TCP acknowledgement. This magnifies the effects of latency and slows overall performance. By coalescing a series of short packets into a single larger packet, and hen compressing both the header and payload, Silver Peak reduces the impact of latency in these types of environments.
Header Compression and Cross-flow Payload Compression
Silver Peak uses crossflow payload and header compression to reduce the amount of HTTP traffic traversing the WAN, improving performance - even on the first transmission.
|
