Speed secrets of application performance

There are various tools to ensure that datacentre applications work better for organisations. However, the true secret to better performance might lie closer to home, within the enterprise.

  • E-Mail
By  Sathya Mithra Ashok Published  August 10, 2008

There are various tools to ensure that datacentre applications work better for organisations. However, the true secret to better performance might lie closer to home, within the enterprise.

All CIOs face it. Applications do not always perform at the highest standards, and detecting the problem and working out the glitches can become an issue, especially when it comes to performance within datacentres. So much so that application performance management (APM) is growing in importance in enterprises worldover.

"Depending on how the datacentre is built, enterprises and end-users may face performance issues, security issues, availability and reliability issues," says Taj El-Khayat, head of enterprise channel group, MEA at Juniper Networks.

Hardware and appliance manufacturers should ensure that the open system architecture is followed, and should liaise with APM suite providers to guarantee that their new releases are supported.

"The most likely application performance issues will reach enterprises in terms of response times and this is something they could face especially within datacentres. Everything else, which includes the analysis going forward and the entire review process across the layers and the modules of the application, follow from the response time equation," states Mahmoud Mounir, software director at HP Middle East.

While some end-users hear about response time problems from the relevant application's users, some others use different measures to keep on top of them.

"I have worked with time stamp solutions before, where any particular application is time-marked when it passes from one layer to the other, or from one stage in the network to another. This helped us in the IT team identify specific problem areas and address them rapidly. This is something we will look to replicate in the Roads and Transport Authority (RTA)," says Indranil Guha, manager of IT infrastructure management at RTA, Dubai.

Layers of trouble

However, response time is just the symptom of a problem and the real problem could turn out to be difficult to locate. The situation is complicated by the various infrastructure elements that applications will have to flow through in a datacentre.

"The hitch is that app developers do not have too much understanding of the infrastructure, and vice verse. The app vendor does specify the requirements and the fine tuning setting of the hardware, on which the application will work best or perform optimally.

But this is often not communicated properly, so apps get put on an infrastructure which is probably not optimal. And the app vendors often do not really know the infrastructure that the end-user will be hosting the app on," says Guha.

"Essentially, at worst you can have an app that is developed by a team of architects in isolation which works fine in a test environment but, when thrown over the wall to the production network, is either slow or unavailable, or inherently insecure. This can mean re-engineering in some cases, which is very expensive, very time-consuming, and ultimately affects user adoption and the bottom line," says Nigel Ashworth, technical director MEA at F5 Networks.

The way a datacentre and networking elements are designed can also pose problems for applications. The more complex the design, the more issues end-users are likely to face.

"APM ultimately does depend on equipment and hardware since the applications are hosted and run on the network and server nodes. There are a multitude of tools to manage infrastructure but the key is to manage these from the perspective of apps and have the ability to correlate and see how failure with an element impacts the app," says Maria Medvedeva, director of technical sales at CA.

Getting the job done

Hardware can often prove to be the cause of an app problem, but an end-user will have to analyse the symptoms first before he can establish whether the problem lies with the hardware or with the app.

"At the end, you want to have a performing app. When you get a problem with the response time, enterprises will need to start the appropriate analysis process. This will involve going into the various layers and looking through them to identify glitches. This is where an appropriate APM tool, with relevant analysis elements can help," says Mounir.

Many vendors, including the likes of CA and HP, have brought out relevant APM tools.

"APM should provide visibility into the infrastructure upon which your apps run - such as your systems and servers, the databases, the authentication servers, the software that these apps are composed of - such as your apps server platforms and your middleware and the app's code itself," says Medvedeva.

3975 days ago

I ran across one that talks about how APM should work: Application Performance Management

4023 days ago

Impressive! this is a good case study. Cyberoam has very effectively managed the security and bandwidth of the network. This firewall is quite unique with it's Identity based security approach.

Add a Comment

Your display name This field is mandatory

Your e-mail address This field is mandatory (Your e-mail address won't be published)

Security code