Businesses globally experience application performance issues, ranging from mildly annoying to severely threatening levels of interference. While not every malfunction costs millions of dollars in lost revenue, even just a brief outage can have long-term effects.
So given the high stakes, why do organizations so often fail when it comes to application performance? There are certainly many factors, but the most frequent causes of application performance errors are usually one of these five issues:
- Systemic firefighting
- Lack of visibility
- Ineffective processes
- Technology and business silos
- Looking for the smoking gun
To overcome these problems and optimize your application performance, you must be aware of them and know how to sidestep potential errors before it affects your usability and productivity. The 5 barriers described below will provide insight into some bad IT habits your organization may be taking part in, and supply methods to counteract them.
Barrier #1: systemic firefighting
A common issue in IT departments is focusing disproportionately on fighting the “fire of the day,” as the cliché goes. While it is crucial to address issues as they arise, it is equally important to avoid neglecting the management of routine IT tasks.
With hours devoted to examining the latest crisis, business, as usual, tends to suffer. This, in turn, negatively impacts the value that your IT team should be contributing as a result.
Systemic firefighting takes a significant toll on workplace productivity, both inside and outside of IT. This misuse of time and resources contributes to the typically high levels of IT staff turn-over and the resulting loss of corporate knowledge. This pattern is frequently blamed on overstretched resources. However, even when spending increases, IT departments still tend to underperform when they fail to effectively balance the hottest problem of the moment with daily maintenance.
One effective method to overcome this barrier is creating and using a matrix of performance issues. Specifically, this matrix should be focused on how each concern impacts the business as a whole and the costs they generate. Having this key information documented in one place will give your IT team a holistic understanding of where your time is best spent. By frequently referring back to this tool, your organization will be better equipped to ensure the effective use of your resources.
Barrier #2: lack of visibility
If you can’t see the issue with your application performance, how can you solve it? There is no hope for fixing a reoccurring problem if you can’t troubleshoot it. The unfortunate truth? This technological blindness is made worse by the disjointed nature of IT departments. Each team investigates using their individual tools, which separately report that there are no errors on their end.
For instance, an attempted application diagnosis often goes something like this: the network teams look for reachability, and the infrastructure teams look for resource constraints. The applications teams are unable to reproduce the issue in their development environment, and the Service Delivery Manager throws their hands up in the air. If a quick answer can’t be provided, technology experts tend to chalk it up to a one-time error, rather than delve into the true roots of the issue. This default laziness can cause a fixable problem to drag on and on.
To achieve better insights into your performance issues, it is essential that your organization quantifies and trends performance metrics, starting with the end user.
Keeping track of these statistics will enable you to observe patterns over time and understand how your performance is doing from all points of view. This will make it easier to clearly pinpoint the origins of a performance malfunction and react accordingly–instead of wasting time and failing to address the issue.
Barrier #3: ineffective processes
Outdated documentation and run books are the source of endless frustration for end users and IT departments. They are typically updated right after the last major IT disruption, and once the situation stabilizes, they are left to the whims of change control and service desk staff.
Unraveling ineffective processes is difficult, as their history often gets lost in the churn of the IT staff that have since moved on. But ultimately, it is as vital to understand the function and reasons for a process as it is to document it.
Some of the most important processes that have proven time and time again to deliver performance value include:
- Application performance triage
- Escalation processes
- Application on-boarding
- Ongoing performance testing
- Quantification of performance
- Historical trending
- Business reporting
These all contribute to improved performance management and if implemented, they can help prevent application performance issues.
Barrier #4: technology silos
In large organizations, cross-silo communication tends to be minimal and even hostile. When I ask what happened following a flurry of activity or outage in a customer’s silo, a team leader will usually reply with, “I heard there was a problem in someone else’s silo”.
Some real-life examples I’ve encountered over the years include:
- “We knew about that, but thought it was too expensive to fix.”
- “We don’t care about 10ms.”
- “It works in development, so it must be someone else’s problem!”
Clearly, the underlying cause is the same. Stakeholders tend to defensively protect their silos rather than work in the interest of the organization as a whole. This segmentation only increases when they can plausibly deny contributing to the IT error.
To overcome this barrier, performance management metrics must be transparent. By shining a light on silos, your IT organization can create an environment of collaboration and joint purpose. This culture shift will discourage any combative instincts and help facilitate effective cooperation between departments.
Barrier #5: the smoking gun
Performance problems are usually not just one issue, but rather involve a number of individual issues contributing to the end effect. When each technology silo sees their contribution to an issue as relatively unimportant, they have no motivation to try to resolve it. In such cases, it becomes far too easy to point the finger somewhere else.
Let’s look at a typical example of an under-performant application in its whole lifecycle. The steps frequently unfold as outlined below:
- The need is identified.
- The business analyst identifies the requirements.
- A feasibility study is performed.
- A prototype or proof-of-concept is produced.
- Production begins to take place.
Before the production phase, the process usually occurs in a controlled environment. These initial failures tend to be dismissed with responses such as, “have you tried turning it off and on again?”
The cumulative effect of each silo’s performance becomes the performance of the whole solution. When a 10% performance degradation occurs in a single silo, it can hurt the overall performance.
In order to counteract this phenomenon of the “smoking gun”, you must prioritize performance issues by identifying their individual importance and the costs required to remediate them. This becomes an iterative process until either you reach the expected performance, or the cost-to-benefit ratio is too high.
By isolating issues and analyzing which ones truly pose an issue to performance success, your organization can optimize your time and protect the general health of your endeavor.
Jumping the hurdle: avoiding performance issues
In sum, many of these barriers stem from disconnected IT departments and a lack of understanding what’s really going on. To optimize your enterprise, you need a unified approach that covers all your bases. Otherwise, your IT organization risks falling short of its targeted application performance.
By Leigh Finch
Published with permission from Riverbed.