Rick Ferguson | ScienceLogic | Cloud Expo Asia | Singapore

Avatar Ben Fower | 21/11/2018

Rick Ferguson, from ScienceLogic, joined us at Cloud Expo Asia to chat about Machine Learning, Big Data and AI Ops. Hosted by Andrew McLean.

Andrew: I have the absolute pleasure now of being joined by Rick Ferguson the vice president of Asia Pacific at ScienceLogic. Welcome.

Rick: Thank you very much it’s good to be here.

Andrew: Yes. It’s quite a show. It’s quite a country. What are ScienceLogic doing here?

Rick: So, we’ve got a booth here. And basically, we are talking to a bunch of prospects and existing customers about primarily the advent of something known as AI Ops.

Andrew: Oh, I like that.

Rick: Good.

Andrew: Right. I know what AI Ops is but what is AI Ops for some of these people who don’t know, start from beginning. What is AI Ops?

Rick: Perfect. So, AI Ops is a term that was coined by Gartner and the premise of AI Ops is that it is a combination of machine learning and big data that when analyzed will produce automated outcomes to minimize manual intervention and manual processes. We think there’s an additional ingredient missing for what we call context but overall, it’s machine learning, Big Data contents and we see ScienceLogic as being the automation engine for AI Ops.

Andrew: Okay. What does that mean for me?

Rick: Okay. So, what it means to you is typically you might have a bunch of people working on faults or problems in your organisation, creating what we would refer to as a ticket on an event, running around trying to get those problems fixed as fast as possible makes absolute sense. The world that we’re moving to is AI. In an ideal world, a predictive world based on historic events and huge data sets. We’re able to predict events occurring before they happen and take remedial action before they happen. So that’s the ideal world. We’re at the stage where we’re able to collect massive amounts of data, you think about all the data you’ve got to collect, information storage, compute, network, security, containers and a lot of this stuff is ephemeral. Now ephemeral is a word that gets thrown around a lot. Is it something you’re comfortable with? Ephemeral means existing for a very short time. So, some of these things are only up for a short time. So, having the capacity to monitor, manage and be aware of these instances and then build context around them and having information around them enables us to take the appropriate action and minimize human interaction.

Andrew: Okay, it’s fascinating stuff. If you spoke to the man on the street. This all sounds very futuristic. I mean I know because I’ve spoken to you guys before but this is happening daily now, isn’t it? Can you give us any examples of something that someone is doing that has used this kind of AI Ops?

Rick: So I think you raise a really good point. I’ll give an example I met with a household name airline very recently and typically what we end up doing for these people is presenting them with topology maps of network infrastructure. So we might show them how a router or a router, depending what country you’re in, connects to a switch, connects to a server and connects to a storage device. That’s all well and good and then we’re able to show if there’s a problem with it going red or if it’s good it’s green or there might be a small problem it’s yellow. That’s all well and good but as a business guy, is that what you’re really interested in? So, the airline what they’re interested in is can my planes take off can my planes land and what will stop those, that’s all they’re interested in. They’re only interested in the impact on the business service. So, customers are moving more towards a prospect, moving more towards a business service for you. So, don’t show me a measure a spider’s web of IT connections. Show me a service dashboard that shows my e-commerce platform is green. That means it’s good. I’ve got a yellow for online branches, I can drill down into that and determine the problem. You know, maybe somewhere in Kowloon some of my branches are offline. And then maybe I’ve got something red that shows I’ve got a catastrophic event. So, I’m able to prioritise and apply resources to where it’s needed most. That make sense?

Andrew: Yeah, it makes perfect sense

Rick: It more of a business view as opposed to the old technology world. It’s solved my problem at the wave of a hand and that’s what you’re here for. The wave of the hand to make magic happen.

Andrew: That’s great. Okay, well we’ve spoken to ScienceLogic in other countries, but I don’t think we spoke to them in Asia, we might have done sorry if we have interviewed you but I’m just looking for trends. What are the current trends you’re seeing? What are the technological trends that you guys are facing at the moment?

Rick: Maybe if I could identify three major pain points that enterprises or organisations are suffering from today and the things we’re trying to address for them. So, the first one as I alluded to earlier is if you’re collating or gathering information from storage networks, from cloud, from your virtual environment, from the security that’s a massive amount of data and just having that data is just useless unless you know what to do with it. Unless you’ve got some context around it. So, the flood of data is a problem for IT organisations. So, we’re trying to help them make sense of all that data Big Data. What does it mean it’s like a telephone directory, you know from the old days? You looked, you needed one name from the telephone directory but you had this massive, massive amount of information. The second thing they’re really struggling with it’s something called cloud sprawl. So, is this something you are familiar with?

Andrew: I’ve never heard of cloud sprawl.

Rick: Okay, I’ll give you a great example. We met with an Enterprise Singapore organisation this week. We sat down with them and we asked them about cloud sprawl. So, the question is tell us about how many instances of public cloud you have deployed? And they weren’t sure how many. They think there were three instances of public cloud employees with the big guys in these users and the AWS. They think there’s one instance of a smaller cloud provider. So, they think they’ve got four. So, the next question was how much is it costing you a month? They didn’t know. So, I said would it be a hundred thousand dollars a month? No, it’s way more than that. $200,000 a month? No, it’s more than that. So, we settled on $250,000 a month. So, they don’t know how many instances of public cloud they’ve got. They don’t know how much it’s costing them and thirdly when I asked them I said what’s the utilisation like on those public cloud instances? Is it two percent meaning that you could spin down the size of the instance? Is it 50% spiking up to a hundred percent, which means you probably need to expand in a bit, and they didn’t know that either. I’m not being critical of the organization because probably most companies are like that. So again, it’s managing that cloud sprawl and also there are funny stories about developers taking a credit card phoning up an AWS spinning up an instance of cloud and doing development on their own instance of public cloud.

Andrew: So, it’s the shadow IT idea.

Rick: 100% and the third thing is something called alert flood which is basically being inundated with notifications alerts and knocks on the door from your boss about instances that you’re probably already aware of. So, its alert flood, it’s too many notifications about stuff you’re already aware about are the three major pain points that we’re seeing.

Andrew: And one last question. Now you mentioned Big Data. I’ve always had a love-hate with big data because I know that companies tend to have, I don’t know if they call it dirty data. Maybe that’s a different idea. But because it’s kind of accumulated in a big box for like a hundred years. Some of it’s almost useless. How do you actually get them to structure that data?

Rick: That’s a fantastic question. So, the way we do it is we use a concept called a data lake and we’re able to add context to the information. You’re so accurately spot on, you should’ve given a presentation for us earlier. It’s so spot-on what you’re saying because it’s the old story of bad data in, bad data out garbage in, garbage out. If you feed an AI system fragmented inaccurate poor data, the outcomes obvious. It’s going to be poor data out. So, we use a concept or principle of a data lake. Where we store and stack information in the correct repositories with context around it so it’s accessible it’s accurate and it can be used in an AI, ML world so we can train that data, learn on that and provide accurate outcomes. So, it’s really good point. The data has to be clean it has to be accurate. It can be a real blockage for AI Ops adoption.

  • © 2019 COMPARE THE CLOUD LTD. All rights reserved.