Slow and inconsistent performance can be a barrier to user adoption and achieving the desired outcomes for a solution. Even small inefficiencies can add up, leading to reduced productivity across the user base.
To get the most out of your enterprise information management (EIM) deployment, you’ll want to ensure that performance is optimal. But some performance issues can be quite challenging to pinpoint, given all the end-to-end interactions and dependencies. This is where following a data-driven methodology can help identify the root cause and the right permanent corrective action.
One of the methods I often lean on (back from my Six Sigma days) is DMAIC, which stands for Define, Measure, Analyze, Improve, Control. The primary goal of DMAIC is the continuous improvement of processes. A user request, for example, is a process. It’s a series of steps: from the user clicking, to the request going across the network, through load balancers, to the application, databases, storage, etc. and back. We want to identify which steps in this process are bottlenecks and mitigate those.
Before diving headfirst into trying to solve an issue, it’s best to start by asking, “What’s the problem we’re trying to solve?” Getting a good definition can help scope and focus the investigation. Then ask, “What is the target performance we’re trying to achieve?” This will set expectations around the end goal, which helps you know when you’ve achieved it.
If we can measure it, we can improve it. We need to move past qualitative assessments like “it feels slow” and instead get accurate measurements. In a perfect world, we would always be able to determine how long each step takes and then drill down for finer details. But the reality is that measurements are difficult to get.
Start with measurements that are readily available. Assess if this captures the bulk of the time you are concerned about, as this will help you if you can then drill down further. If the problem is outside this measurement, expand your focus to measure further up the stack. Let data guide the way.
You can avoid guesswork and random shots in the dark by using data to home in on the primary bottlenecks and their root causes. One way to do this is by temporarily reducing variables where possible. This helps to rule things in or out. For example, if you suspect the network might be the issue, simulate an end-user test directly from the server; this eliminates the load balancer as a potential issue. Doing this will either confirm or disprove your hypothesis and point to the right direction for further analysis.
Once you’ve found the root cause, you’ll need to figure out if and how it can be improved. In some cases, this might involve a simple configuration change or tuning. At other times, it might require software patching, an upgrade or even infrastructure enhancements. Be open to non-technical solutions, too, which can lead to the best user experience. Is the business process making the best use of the technology or should it also be optimized?
Once you’ve implemented a solution, make sure it not only achieves the performance levels you were hoping for but also ensures those gains are sustainable over the long term. This means ensuring that both good performance monitoring and change control are in place. Another thing to ask: Should any measurements that were made during earlier phases now become part of the standard monitoring used to track the system?
Like onions (and ogres), performance has layers. So it’s often the case that clearing one bottleneck will simply lead to the next one. This means you’ll need to keep repeating the DMAIC process to continuously improve.
Optimizing performance with OpenText
Performance tuning can be daunting. But you don’t need to go it alone. Regardless of where you run your EIM platform, OpenText™ can help ensure that your end users see optimal performance. If you’ve chosen the OpenText™ Cloud, we take care of performance within the hosted infrastructure.
OpenText Professional Services can also assist with performance—whether that’s on-premises, in the public cloud or anywhere in between. With years of experience in running production systems and overcoming performance hurdles, OpenText can provide short engagements through our Health Check and Performance Optimization services, as well as ongoing help through our Optimize Assist Program. Contact us for more information.