The Next Level of Application Performance Management (APM) in the Cloud

Cloud Management

The Next Level of Application Performance Management (APM) in the Cloud

By TMCnet Special Guest
Michael Kopp, Compuware
  |  January 25, 2013

This article originally appeared in the Q1 -2013 edition of Cloud Computing Magazine.

Businesses expect many benefits when they move to the cloud, including greater business agility, significant cost-savings and of course, increased profitability. Careful application performance management (APM (News - Alert)) can be the key to achieving these benefits, in more ways than one.

Recently, there has been a lot of discussion around the importance of APM in the public cloud, particularly as it pertains to ensuring strong end-user application experiences. The cloud is opaque, meaning that cloud customers often have little insight into the inner workings and capacity management decisions of their chosen cloud service provider. The only way to know for sure that your end users are having a fast, reliable experience with your application is to measure this performance from the true end-user perspective on the “other side” of the cloud, and use these insights as a basis for establishing and upholding application performance-focused SLAs. In addition, you need end-to-end, deep-dive diagnostics to help you identify the source of performance problems, to determine if a problem is based in the cloud, your own data center or another element in the application delivery chain.

Many businesses have made significant progress in using APM to ensure a high level of cloud application performance. They may have reached a point that their applications are fast enough, but this is no time to stop doing APM in the cloud. There’s tremendous benefit to be gained by having a better understanding of the inner workings of cloud-based applications. The next level of APM in the cloud is all about optimizing the cost structure of your cloud-based application, which, just like application performance, has a direct impact on cloud return-on-investment.

Put another way, at a certain point, it’s no longer just about making applications faster – it’s also about making applications more efficient from a cost perspective, and this needs to be taken into consideration at every step of the application development process. Consider a search function – you need to optimize it so that it delivers better results and is not executed five or more times by every user. This can be considered functional optimization, but also lowers the operational cost, because in the cloud every transaction has a dollar value attached to it.

Making less database access calls per search transaction, while not making the search faster at all, can save money. This is because most cloud providers charge on a per operation – e.g. per SQL – basis. Thus optimizing the number of SQLs might be more cost effective than saving CPU! In this approach, businesses completely sidestep the question of resource optimization and go straight to where it matters – cost optimization. The fact is, we don’t really care about resource usage in a public cloud at all. We care about the true end-result – application SLAs and about cost effectiveness.

Another example is the purchase function. Are there any features that consume a lot of resources in the cloud – product tours or images, for example, that end-users spend a lot of time on? This could be driving up your costs in the cloud. Businesses need to know the cost structure of a transaction and how much revenue it generates in order to set priorities, and so transactions leading up to conversion can be optimized both functionally and cost-wise.

For many companies, one of the primary perceived advantages of moving to the public cloud is elastic scaling. Elastic scaling avoids the old way of capacity planning and big expenses upfront; instead businesses can increase the size of their environments as their load increases. But elastic scaling also has a disadvantage – over-consuming planned capacity can happen easily since there is no hard limit, which often leads to exceeding cost-estimations. It’s therefore critical to directly understand how end users interact with an application (where APM in the form of end-user experience monitoring comes in) and how the application handles the load. Not having this information is operational blindness and comparable to simply passing on the company credit card. 

In summary, end-to-end user experience management enables us to understand our users’ behavior and how performance affects conversion rates and our business. But in a public cloud this is only half of the APM story. Only if we can accomplish both – satisfy end-user expectations for fast, reliable applications while keeping costs down – can we be successful in the cloud and improve our business performance as a result of using the cloud.

Michael Kopp is the technology strategist of Compuware (News - Alert).




Edited by Brooke Neuman
blog comments powered by Disqus