Acceleration

 

This section on Acceleration describes the Ipanema System features that reduce the response time of applications over the WAN.

 

Acceleration features allow to:

 

 

 

 

 

 

Accelerate while protecting critical application performance

 

The need for Acceleration should not be mistaken with the need for Optimization. Optimization is a mandatory technology to ensure that the response time of critical applications is kept minimized. When congestion occurs, the size of router FIFO buffers naturally increase, the transit delays get higher and packets start getting lost, thus degrading application performance.

 

Moreover, many acceleration technologies rely on the ability to go beyond what network protocols were designed to do to use network resources. For example, they allow flows to “fill the pipe” in situations where network protocols prevent them from doing so. Without a proper combination with Optimization technologies, most Acceleration techniques will actually worsen the delivered application quality because they create or worsen congestion situations.

 

Ipanema is unique in its tight coupling of Acceleration with its other core feature sets of Visibility and Optimization.  That ensures our Acceleration is always controlled, and used in the service of achieving the business-driven application performance objectives the enterprise has defined. With Ipanema, Acceleration will never be a source of performance degradation.

 

Implement both a strategic and tactical approach to Acceleration

 

There are 3 opportunities for Acceleration:

 

  • The TCP bottleneck: because of its “session by session” design, TCP is not able to make use of all available network resources.

 

  • The bandwidth bottleneck: the performance of many applications is related to the available bandwidth. If it is higher, then the response time of the application will be lower.

 

  • The application protocol bottleneck: some applications are not designed for the WAN. Their application protocol is chatty, relying on too high a ratio of protocol exchanges to transmitted data. When the network delay is increases, the performance degrades.

 

The first two of these bottlenecks are more fundamental than the third.  In fact, application protocol bottlenecks should really be considered as a tactical issue by savvy enterprises. All application protocol bottlenecks, without exception, can be removed simply by updating the design of the application to take into account WAN style network delays. This is what Microsoft has been doing for many years. Exchange 2003, for example, exhibits vastly improved performance over its predecessors when used over the WAN. CIFS, a protocol known for it very bad WAN performance has been completely redesigned to solve WAN performance issues as part of the Windows Vista and Longhorn operating system releases.

 

Ipanema is addressing all 3 opportunities for Acceleration to cover strategic as well as tactical needs.

 

 

 


Unleash TCP performance without branch devices using Tele-Acceleration

 

The two key TCP performance limitations are its “slow-start” phases and the size of its “window” resulting in the inability to use all the available bandwidth on longer-delay, larger-capacity connections (the “bandwidth delay product problem”).

 

The “slow-start” is the TCP mechanism that tries to discover what the available bandwidth is for each session. This mechanism is used by every individual session and relies on a progressive increase of the session throughput until it reaches the point where the link is congested. It assumes then that it has found the maximum available bandwidth. While it is slowly increasing how much bandwidth it is using, it is by definition not using all the bandwidth it could be, and thus is under-performing during that phase of the connection.  Some bandwidth is unused leading to a response time that is higher than it could be.

 

The Ipanema System leverages its knowledge of the optimum throughput for each session that is gained from the Visibility and Optimization features. With a “local acknowledge” mechanism in the ip|engine located near the source of traffic, Ipanema Tele-Acceleration immediately sets each session to its optimum bandwidth, leading to an important improvement of the response time of numerous applications, including those based on HTTP or HTTPS.

 

The “bandwidth delay product problem” describes TCP’s inability to use all of the available bandwidth when the number of simultaneous sessions is low and when the product of bandwidth and delay is large compared to the TCP window size (TCP window size depends on TCP implementations on the client and the server).  For example, a backup performed at night with a single TCP session on a high-delay/high-bandwidth link will use only a fraction of the available bandwidth leading to a backup duration that is much longer than it could be.

 

Ipanema addresses this issue by creating a virtually unlimited “window” size through controlling the packet transmission rate at the source.
Unlike any other solution on the market, Ipanema is delivering TCP acceleration features without the need for a device in branches. Devices are only required at the source of the application flows.

 

 

 


Locally cache and compress data using Multi-Level Redundancy Elimination

 

Ipanema's Multi-Level Redundancy Elimination is an advanced technology that reduces the amount of data transmitted over the network by compressing and locally caching traffic patterns in a cache in ip|engines located at branch offices.

 

The process frees network resources that are then immediately made available to applications (starting with the critical applications). As a result of the creation of this “virtual” bandwidth, applications perceive a network link as if it were much bigger than it is, leading to a reduction of response time.

 

When using RAM caches, Multi-Level Redundancy Elimination is able to compress and cache any type of application patterns – even those based on UDP. When using both RAM and disk caches,  Multi-Level Redundancy Elimination is able to compress and cache the patterns of very large files transported over TCP and keep them over long periods.

 

 

 

 

Transparently accelerate legacy applications using Intelligent Protocol Transformation

 

Even though many CIFS performance issues are solved in Windows Vista and Longhorn, enterprises are not yet fully deploying this latest Microsoft technology. Those who need to access CIFS-based file shares over the WAN, after a storage consolidation phase for example, require a solution to improve the response time of CIFS operations.

 

Ipanema CIFS acceleration is based on the transformation CIFS protocol exchanges in order to minimize the protocol's vulnerability to delay and thus to deliver an appropriate response time when accessing remote file shares with CIFS over the WAN.

 

Optimizations on the protocol are achieved using the Intelligent Protocol Transformation technology that provides a generic framework for optimizing application protocol patterns that are eligible for prediction, aggregation or translation leading to a minimized application response time.

 

 

 

   Top