Problem solve Get help with specific problems with your technologies, process and projects.

Improving cloud computing performance: A path for revenue growth

Instead of seeing customers' cloud-computing-performance concerns as a barrier, providers should regard it as an opportunity to offer application acceleration services.

Editor's note: This is the first part of a three-part tip series on how service providers can optimize cloud computing performance.

When customers migrate critical applications to the cloud, they expect cloud computing performance to be impeccable. Yet when enterprises relinquish ownership of and control over the underlying infrastructure, it is hard for them to feel confident that applications will perform well enough to make users productive and happy.

Enterprises have high expectations for cloud computing performance with business-critical applications, and their lack of confidence continues to be a growing barrier to adoption. Most cloud providers offer service-level agreements (SLAs) for availability, but they are silent about performance assurance. If an application is available but poor response time has users twiddling their thumbs, the application might as well be down. Ensuring this doesn't happen spells opportunity for cloud providers.

Forces that degrade cloud computing performance

Many factors contribute to slow application and cloud computing performance. Chief among them are long distances, high volumes of client-server interactions, big payloads, insufficient bandwidth, network congestion and server bottlenecks. 

A big culprit for poor application response time is latency, the round-trip time between the user and the application server. Limits imposed by the speed of light cause latency to grow as the distance between user and cloud server increases, and when you multiply latency by the number of client-server interactions, application response time can quickly tank. The following equation by application performance expert Peter Sevcik, president of NetForecast, illustrates the many threats to cloud computing performance:

Application performance equation

R is the task response time, which is the elapsed time (in seconds) between a user action (e.g., mouse click, enter, return) and the system response (client, network, server), so the user can proceed with the process.

Payload is information content (in bytes) that must be delivered to and from the user's device.

Bandwidth is the minimum bandwidth (in bits per second) across all network links between the user and the application server. The slowest link is typically the user's access line to the network. Usable link bandwidth may be reduced by traffic conflicts (congestion) and protocol inefficiency (e.g., small TCP window size).

AppTurns are the application client-server software interactions (also known as "turn count") needed to generate a user-level system response or task. Turns do not include two-way TCP interactions (e.g., ACKs). 

RTT is the round-trip time (in seconds) between the user and the application server.

Cs (Compute Server) is the total processing time (seconds) required by the server(s).

Cc (Compute Client) is the total processing time (seconds) required by the client device.

Cloud computing performance: Symmetrical vs. asymmetrical solutions

Fortunately for cloud providers, there is a growing body of techniques -- each addressing one or more elements in the above equation -- that can be used to improve end-user response time. Some of these techniques are incorporated into symmetrical, or dual-ended, deployments while others are asymmetrical, or single-ended. They can often be deployed as hardware or software.

More on cloud computing performance

Cloud computing performance FAQ: What makes the cloud go?

Providers must offer cloud performance testing for customer apps

Application Performance Management as a Service: What does it take?

Symmetrical-optimization technologies, such as most wide area network (WAN) optimization products, have one component that is deployed near the cloud servers and another component located near users or on the user's device. These products work on all networked applications and include techniques like TCP optimization to improve throughput; dynamic compression and data deduplication to reduce payload; remote file caching to reduce round-trip time and payload; and application-turn reduction to lower the number of round-trip requests an application may make by gathering data into a single transaction. 

Like their dual-ended counterparts, asymmetrical-optimization technologies (commonly marketed as content-optimization solutions) are also deployed near cloud servers and accelerate a subset of traffic, including Web traffic. They are usually deployed in a data center as software or a standalone appliance.

Because the user's endpoint must be able to understand any modifications that an asymmetrical technology makes to the data, the data-center device must communicate with the client software (usually the Web browser) that has to make sense of these modifications.

Asymmetrical-optimization products include compression to reduce payload; local caching and SSL offload to speed server compute time; HTML transformation, which instructs the browser to retrieve content more efficiently, such as consolidating individual requests for page elements; TCP offload to funnel traffic from many connections into a single persistent TCP connection in the server to limit CPU consumption; and a variety of emerging techniques that accelerate content more efficiently. 

Check out the rest of this series

Better cloud application response time: An overview of partner options

Cloud APM services win customer trust by monitoring app performance

Enterprises care more and more about the performance of their networked applications. If providers develop the right set of services to enhance cloud computing performance, they will create an avenue for business growth. In the same vein, they should also add cloud computing performance measurement and monitoring services.

Ultimately, a provider should be able to implement SLAs that guarantee cloud computing performance and to grow its performance-related service portfolio to enhance the bottom line and reduce churn.

Coming up next: In the second part of this three-part tip series on optimizing cloud computing performance, Rebecca Wetzel lays out a vendor comparison for cloud providers:  Better cloud application response time: An overview of partner options.

About the author: Rebecca Wetzel is a principal with NetForecast and also is president of the marketing consulting firm, Wetzel Consulting LLC. She provides data communications industry insight and helps vendors and service providers develop successful marketing strategies.

This was last published in April 2012

Dig Deeper on Telecommunication networking