Increasingly, larger enterprises are using cloud computing for their mission-critical applications. With business success tied closely to these applications, performance has become an area of growing importance.
But cloud application performance touches on a complex web of interactions that often are difficult to gauge. A handful of third-party tool vendors focused on cloud app performance have emerged to address such challenges; however, these products can present a series of tradeoffs, rather than solutions.
Because cloud app performance tools are complex, they are geared more toward large enterprises than small businesses.
Radar Service from Cedexis Inc. measures response time from more than 1.7 million end nodes globally. The tool vendor offers a free service, supported by an open source model that examines major cloud service providers; Cedexis also offers a paid version for businesses that want to more closely monitor their own cloud connections.
CloudSpeedTest from CloudHarmony Inc. allows enterprises to benchmark the performance of applications delivered by multiple cloud providers. For example, it can monitor one cloud installment that supports servers, a second that hosts application storage and a third that handles the content delivery network. The CloudHarmony service relies on a variety of benchmarks developed by the Phoronix open source community.
CloudSleuth from CompuWare Inc. monitors the application latency for leading cloud services, such as Amazon Web Services, Liquid Web's Storm on Demand and Rackspace managed cloud services. To gauge performance, CloudSleuth monitors about 200 different performance elements and conducts approximately 3,000 tests daily.
Cloud app performance tools don’t show it all
These cloud performance tools can be helpful to give enterprise IT an idea of how their applications are behaving, but they don’t tell them everything.
First, service visibility is limited. The tools can’t monitor every item and every application at every second, so they sit at different places on the network and take various measurements to extrapolate cloud app performance metrics.
One tool may look at end-user response time, while another examines the number of packets pushed out from a server. Tools may run tests every 15 minutes or every minute. Therefore, finding apples-to-apples comparisons can be difficult.
And, even limiting what performance metrics the tools test, results can become extremely complex. To make sense of this, third-party cloud tool vendors need to build sophisticated analytic features into their systems. The process is more art (best guess) than science (definite correlation).
Consequently, tool vendors don't guarantee a customer's application will perform in a certain manner within your cloud provider's infrastructure. Instead, the benchmarks merely provide guidelines. In the end, enterprise IT is responsible for fine-tuning these services to meet their company’s expectations and performance criteria.
In addition, cloud networks change dynamically and quite drastically depending on how much information flows over the Internet. As a result, it can be difficult to mimic the circumstances leading to a performance problem. Consequently, troubleshooting can eat up a lot of IT resources and drag out for days, weeks, or even months.
Because cloud app performance tools are complex, they are geared more toward large enterprises than small businesses. The monitoring tools have various usage-based pricing models that can be difficult for some smaller companies to justify.
As cloud computing use continues to emerge within the enterprise, IT teams understand the need to carefully monitor application performance. While various cloud app performance tools and services are available, they're not a panacea; more work needs to be done before these tools offer all enterprises the visibility they require.
About the author
Paul Korzeniowski is a freelance writer who specializes in cloud computing issues. He is based in Sudbury, MA and can be reached at email@example.com.
Dig deeper on Network and application performance in the cloud