Our story starts on what could be just any day at the office. Someone has just gone to get coffee, when the door burst open. It’s the CFO complaining about the cost of keep over-running your 4-hour rolling average MSU utilization. You’re about to explain when an angry department head pushes his way into your office angrily wanting to know what it will take for you to keep to your service level agreement with his team. But before you can start to calm him down, the phones start ringing with your mainframe users complaining that everything is running so slow.
If this sounds like it could happen in your office, what you need is training in Mindfulness and how to stay calm under pressure, or, better, something that can let you know when these problems are about to happen – so they can be dealt with immediately. What you need is a baseline, so you know when things are running normally, and something to let you know when things are moving away from those baseline numbers.
An always-on software expert is what you need – one that can understand normal system behavior, and, with that information, recognize anomalies and predict when troubles are coming. You want it to be able to consolidate and unify performance data from multiple systems and LPARs; to analyze and learn from the data to make quick useful judgments. You want it to be able to predict and alert you when demand exceeds capacity. And you want it to save all the necessary diagnostic data in real-time as the problem occurs. And, of course, you want it to alert those system experts who are best able to fix the root-cause of the problem.
ConicIT can do all these things. It is not another ordinary monitoring tool, nor does it simply provide just a consolidated view of performance data. It is an expert system that collects, cleanses, and interprets monitoring data from the z/OS monitoring infrastructure. It uses mathematical modelling and machine learning to head off performance problems before they become business expenses.
Because the product draws on many different sources, it builds a nuanced profile of your system based on dynamic thresholds that eliminate false alerts and reveal latent issues. A proprietary data-cleansing component eliminates noise and normal fluctuations and only reports meaningful deviations in real time. By applying a composite, multi-expert prediction algorithm to any anomaly, ConicIT determines a coherent causal chain, identifies the root of the emergent problem, and enables the right people from your team to fix it before service is disrupted in any way.
ConicIT runs 24×7 outside of z/OS, on a Linux server, where it makes use of your existing z/OS monitoring software. When you start using it, you can say goodbye to there being a time lag between when a problem occurs – i.e. the time when meaningful, analyzable diagnostic data are available – and problem detection, i.e. the time when the monitoring data only shows the symptoms, not the cause. As we’ve all experienced, that time lag is a fundamental weakness of common performance-monitors. You can also say goodbye to staff wasting time trying to recreate a problem so they can see what actually happened. You’ll have the data you need.
The benefits of using ConicIT (your friendly neighborhood online expert) are that it reduces the costs associated with overages on sub-capacity pricing; it prevents catastrophic service outages and recurrent issues; it boosts user productivity and customer satisfaction levels; it reduces MTTR (Mean Time To Recover) and increases MTBF (Mean Time Between Failures); it learns from both normal and abnormal states and dramatically improves operational efficiencies; it reduces penalties caused by SLA non-compliance; it can help to rebuild trust and confidence in IT service and support; and it can reduce your costs.