Tuesday, May 21, 2013

Performance Management in the Cloud: Challenges & Solutions



 



By Erin Harrison, Executive Editor, Cloud Computing
 
In today’s complex networking environment, IT operations teams spend the vast majority of their time trying to locate relevant information, or sifting through vast amount of unusable data when they look for network problems. This job is made more complex when the network is distributed among in-house, remote and cloud infrastructure – it’s like following a moving target.

As technology changes continue to accelerate, service providers are seeking new ways to manage more complex, high-bandwidth, IP-based data services. A consistent challenge for IT managers and administrators is to proactively monitor performance across the cloud, identify threats and decrease any impact on the end customers.


Simply moving your applications, or part of your applications in a hybrid cloud setting, doesn’t mean you move the responsibility to manage those applications, explains Vic Nyman, COO and co-founder of BlueStripe Software.

“You still have to manage service levels and solve application performance issues. IT operations teams must always monitor their applications and be prepared to determine the cause of performance or availability issues,” Nyman says. “Gathering performance data across a dynamic real-time cloud environment is one place where transaction monitoring tools have a big edge over conventional server and application tools, which tend to be blocked by cloud systems.”

According to Frost & Sullivan’s (News - Alert) definition, the network performance management market consists of solutions to manage, monitor, resolve faults and control changes on the network infrastructure. These solutions are designed to track network traffic gathering information for the needs of statistics collection as well as measuring and optimizing performance for network elements.

There are several kinds of network performance tools available on the market today. For example, conventional network and server monitoring tools focus on resources – including available CPU, memory, bandwidth, etc. – as opposed to service levels. 

“This can lead to situations where all performance lights are green, but customers are experiencing problems. Conversely, just measuring end-user response times without an understanding of where user transactions go in the data center leaves a large management gap, too,” Nyman explains.

As more applications are moved into the cloud, an end-to-end transaction monitoring solution is required. “For today’s complex applications, especially those incorporating some form of cloud architecture, it’s an absolute requirement to have an end-to-end transaction monitoring solution that sees the real-time connections between servers and processes,” he adds.

BYOD and SLAs

The Bring Your Own Device (BYOD) trend presents yet another cloud-related hurdle for organizations. The rapid proliferation of consumer mobile devices is changing the traditional IT environment in enterprises, as 90 percent of enterprises have already deployed mobile devices, with smartphones being most widely deployed, according to a recent survey by Gartner (News - Alert).

“BYOD adds new requirements to every single application. Those extra requirements create additional complexity and opportunities on top of the already complex mission critical applications. It’s application complexity that creates performance management roadblocks for IT operations teams,” says Nyman.

Performance monitoring tools alone do not allow businesses to better meet SLAs and provide continuous operation of services. The primary function of an ideal performance management system is to optimize the use of the network and applications so as to provide a consistent and predictable level of service.

“Just having tools won’t help you meet your SLAs. It’s important to match tools with the performance goals. For SLA improvements, tools must understand how components (either in the data center or in the cloud) impact user transactions,” Nyman explains. “IT operations teams need to know not just when something fails, but where the failures occur and why. Once this data is understood, IT operations can quickly take care of any service level issue by tracking it right to the problem component.”

Taking an End-to-End Approach

Another challenge around network performance monitoring is that many IT managers lack confidence in their own processes to find network performance problems before end users report issues. 

In fact, a recent BlueStripe Software survey showed a disconnect between goals and the processes designed to achieve those goals. More than 80 percent of enterprise-scale companies never solve a quarter of their application problems.



“This seems shocking at first, but with today’s ultra-complex applications, it’s understandable,” explains Nyman. “There’s really no way for IT operations managers and team members to understand all the dependencies between servers, processes, applications, and back-end systems. The platform-focused approach most management tools take don’t alleviate that concern, adding an additional requirement of platform expertise just to use the tool.”
That’s why end-to-end transaction monitoring tools are so critical to IT operations teams today, adds Nyman. With massively complex and dynamic architectures, it’s the only way for IT managers to see how any given server or system is impacting end users.

For service providers and other businesses, there are several benefits of working with a network performance vendor. With so much of today’s business success tied directly to IT systems, having an effective service level management solution is critical to maintaining business processes, keeping customers happy, and achieving revenue goals.

“I know that as far as BlueStripe is concerned, we provide more than just the tools to monitor transactions. We also bring broad expertise,” says Nyman. “What looks like a new problem – say, for example, a failure in messaging system integration or chronic slowdowns in Websphere MQ – to you is probably something that we deal with every day. In that way, not only can we help you achieve best in class monitoring systems, we can help you quickly eliminate common issues – that’s especially important with the complex systems in play today.”

No comments:

Post a Comment