The Chicago Police Department turbocharged their detective work by putting in action an advanced pattern-recognition software system. Created by computer scientists at DePaul University, Classification System for Serial Criminal Patterns (CSSCP) uses artificial intelligence to cull massive amounts of data.

Introscope admins at a major credit bureau were frustrated over the typical 90 minutes they spent on conference calls just trying to figure out where to start the troubleshooting process.  They gave Prelert a chance by feeding it data from a recent incident where CPU utilization for the Websphere JVM process pegged at 100% for 30 minutes resulting in the restart of multiple services for a critical production app. Prelert’s intuitive dashboard (shown below) allowed the admins to go back in time to view anomalous activities related to the CPU utilization spike.  Prelert identified a web service that took ...

Researchers at MIT have demonstrated the power of leveraging self-learning AI to allow computers to get better at playing video games.

You can use “Google Correlate“ to see if arbitrary items correlate well. One of the examples is a correlation between the Federal Reserve’s balance sheet and the number of Google searches on “nausea remedies”...

In episode 3 of “The Code” (http://www.bbc.co.uk/programmes/b0137xfr), TV mathematician Professor Marcus du Sautoy demonstrated how accurate a group can be at making predictions. He filled a large jar with jelly beans, then asked 160 people how many they thought it contained. The individual estimates ranged from 400 to 50000! The correct answer was 4510, so [...]

Last week I was at a prospective customer in a newly renovated office and travelled up to operations floor in a new ‘Destination Control Elevator’. These are the ones where you have to press a button to indicate your destination floor before boarding. Inside the elevator there is only a stop button as it has been already programmed to go to your selected floor. To be honest it was slightly unnerving, as you’re in a sealed metal box and have to trust the programming of the elevator system to take you to the correct destination. The benefit of these ...

Each of the CIOs believed that their organizations certainly had the monitoring data required to diagnose the root cause of application/service problems. It was interesting to watch as their technical advisors shifted in their seats, cleared their throats and politely opined that ‘actually, this is a consistent problem’. One can hardly blame a CIO for believing this myth since, after all, they’ve spent millions on monitoring systems. But the CTO’s, VPs of Operations and Application Support owners know the truth. In practice they only turn on ‘deep’ data collection when a problem occurs and the cause is ...

I recently watched the BBC documentary “The Joy of Stats” (http://www.bbc.co.uk/programmes/b00wgq0l) with Professor Hans Rosling and thoroughly recommend this and his other lectures (http://youtu.be/hVimVzgtD6w). Professor Rosling has the very rare skill of presenting technical content with humour and great enthusiasm, and the documentary describes some of the phenomenal volumes of data that are being generated [...]

A recent survey of 100+ IT executives in mid to large enterprises shows concern over the just how far you can trust your cloud vendor. 65% of respondents said they planned to monitor the performance of applications they’ve put in the cloud with their own IT operations management tools. They (64%) also thought they’d need [...]

A well-known issue that plagues application performance management systems is what threshold values to set for alerts? For instance, should alerts be raised if CPU is above 90%, or Average Response Time is above 500ms?

As follow up on my Data Driven Revolution post , let’s consider the difficulty of pattern detection analysis given the huge volumes of data it is now possible to collect.  Specifically how self-learning technologies (3rd generation machine learning) offer the best promise given the complexity of the today’s target systems. I will rely further on [...]

I recently watched Professor Chris Bishop’s IET/BCS Turing Lecture on “Embracing Uncertainty: The new machine intelligence” (http://scpro.streamuk.com/uk/player/Default.aspx?wid=7739). For those of you unfamiliar with his lectures and books, I thoroughly recommend them as a great source of knowledge and summaries of the current status of research into machine learning. His introduction to the lecture describes a [...]

We recently requested feedback on a number of Linked In groups concerning the ability of our self-learning analytics to accelerate ITIL Incident and Change Management processes. We appreciate any feedback you leave as a comment. Thanks Share on Facebook Tweet This Post

So last week was a bad week for Windows Phone 7 users as Microsoft experienced two major outages in a matter of days. On May 4, 5 users were unable to buy or download apps or access Xbox Live for 10 hours.  According to Microsoft the ‘root cause’ was identified in a matter of hours [...]

How can I diagnose application performance problems when the data is no longer available? One great thing about writing software products is when customers start finding new use cases for the existing product. An example is a recent customer who is exporting data from our causality view and attaching the data to trouble tickets. To [...]