Subscribe to our 0800-DEVOPS Newsletter

    Get in touch

    Contact us for all inquiries regarding services and general information






      Get in touch

      Contact us for all inquiries regarding services and general information






        Use the form below to apply for course





          Get in touch

          Contact us for all inquiries regarding services and general information






          Blog

          Artifical Intelligence in IT – just another hype or…?

          27.04.2021

          A lot of hype and a lot of choices

          Being in the industry for the last 15 years (time really flies!) I was able to witness numerous trendy buzzwords, hypes, new standard and architectural patterns being put out there as something you (our valued Clients) ‘must have’. Sometimes it was a tech trend (piece of new technology or concept) and sometimes a brand new methodology.

          Let’s just remember a couple of them:

          • Three tiered architectures, SOA, EJB, BPM, Microservices and Cloud Native, JSP, JS
          • Waterfall, Iterative development/ RUP, Agile
          • Relational Database, NoSQL, Hadoop, BigData, IoT, Blockchain

           

          Let’s not forget – these really are high-level terms; underneath, there are literally hundreds of tools and concepts and, eventually, choices to be made.

          By looking at this list, I can recall that every single one of them had its pros and cons, with much more pros. All major vendors were pushing them to the market and, over time, they became a standard, whether you like it or not. I would argue that most of the enterprise Clients still have all of the above mentioned trends somewhere in their organization – some of the components running happily in production and bringing value, and some of them being a constant headache or bottleneck.

          Picking the right one

          At CROZ, we like to think that we owe our success to the ability of picking the right tech trend at the right time and riding on that tech wave until the next one. Of course the general idea is to solve our Clients problems by using the right tool. In other words, we were many times on the forefront of proposing new solutions/concepts to our Clients.

          I won’t go into much details and explain how we pick the right ones, but it comes down to couple of factors (this is obviously very simplified):

          • The initiative almost always comes from our tech experts
          • Then we need some time to assess the technology/concept by experimenting by ourselves and with some trusted clients
          • And only then we decide – OK, this is cool, our tech guys love it, it brings value to the Clients, let’s do this more…
          • If it all works – it becomes part of the strategic area and Client offering

          AI – is it a magical computer based intelligence or just another way to program?

          Even if you are not in the tech industry, you must have heard about AI. Which is kind of interesting… For example – my parents, wife or brother in law have not heard about SOA or BPM, or even Cloud Native as a concept. But they sure have heard about AI.

          We all know about infamous Skynet from Terminator 2 and evil robots taking over the world… And the date in the movie when this war between man and AI powered machines happens is August 29, 1997! The AI was actually very frequently used in movie industry and pop culture for a long time. I personally don’t believe that robots will become evil, self-conscious and start launching nukes on people, but it’s definitely time to take this AI beast into real consideration. Is it just a buzz or can we actually create some value to our Clients by using the power of AI?

          We started looking into this area (Machine Learning to be more specific) a couple of years ago when we developed our own chat bot. This was actually the first opportunity for me to see ML in action playing a human who chats with me. And I must say I was positively surprised. At certain points I actually had a feeling that a real person was somewhere behind the keyboard.

          When my non-IT friends ask me what Machine Learning is, I usually explain it like this: “Well, you provide large amount of historical data and the program is able to learn from it in a way that it can classify future inputs based on that historical data”. For example, if you load 10 million customer emails and clearly state to the program which of those are complaints, new service requests, etc., the program will be able to recognize and classify all (or most) future emails itself”. And voila we can already see the benefits in terms of savings and even improved customer experience.

          Employ AI to take care of your logs and events

          Fast forward to today. We started working on some real use cases that can be applied to almost any enterprise. One of those is so called AIOps.

          15 – 20 years ago, a main issue in Operations was lack of information (app logs, system logs, etc.) from different sources and various systems. Mind you, the IT environments were somewhat simpler than today. Still, we were concerned how to:

          • collect enough information from client-server, Java and .NET web applications
          • collect enough information from networks, various OS-s (mainframe, Solaris, HP UX, SCO Unix, W2K…)
          • correlate the collected data into meaningful information
          • Create basic automation (e-mail, SMS, corrective actions)
          • show the data as “Business view” (one picture for all IT silos)

          Today, we have a different challenge. We can gather huge amount of data from huge amount of different systems and use a lot of advanced tools to visualize that data. So, the concerns are a bit different and they come down to how we can:

          • get only the important data (too much information)
          • react before something happens (in agile environment) and integrate with incident mgmt.
          • maintain the alert thresholds in dynamic environment (what is normal?)
          • see the real service levels that users see

          In short – today we have too much information and it is becoming very hard to determine what is actually important for us to prevent or solve an ongoing issue in production.

          This is where our before mentioned Machine Learning comes to play. By using the tools and techniques of Machine Learning we can (of course not manually but automatically) do things like:

          • Learn the normal system behavior and set alert thresholds automatically, by going through large amount of data
          • Anomaly detection, including business impact
          • Detect recurring problems, issue clustering (grouping) à RCA, and enable self-healing
          • Intelligent Incident management (improving collaboration)

          The above mentioned benefits are primarily related to monitoring, but it’s worth to add that we can employ AI/ML techniques to improve and automate engagement (Task Automation, Knowledge Management, Change Risk Analysis….) and automate actions (Scripts, Run Books, ARA).

          So, to conclude – we at CROZ strongly believe that AI has a great potential to really bring a positive impact on operations (and many other fields of course) if approached with right motivation, goals and tools.

          Stay tuned for more on this topic!

          CONTACT

          Get in touch

          Contact us