AITSM Show - Episode 4 - Special Interview With Girish Gadage
How To Make A Good Automation Initiative Even Better?
What Is Important & What Is Not?
Five Tips To Get More Value Out Of Automation.
AITSM Show - Episode 4 - Special Interview With Girish Gadage
How To Make A Good Automation Initiative Even Better? What Is Important & What Is Not? Five Tips To Get More Value Out Of Automation.
Automation always starts with zeal but dies out after a few months because either they don't see enough value in it or lots of effort/rework is needed to move the needle.
We'd our pertinent questions like - In Automation, What Are The Important Things to Look For & What Is Not? How To Get More Value Out Of Automation. How To Make A Good Automation Initiative Even Better?
We asked it from Girish Gadage, Global Head, Tooling & Architectures at ATOS. With plenty of examples and anecdotes from his rich and diverse career, we'd learn a lot from him.
Don't miss the unique set of five tips shared from the rich experience by Girish to make a good automation initiative better or even the best.
"The best way of doing automation is to define a business objective first. What do you want to achieve from automation? Then you look at the technologies on hand. Don't go out of the way, just because some technology is the leading technology in the market."
Learn and Grow, With AITSM Show!!!
Anand Tambey: Welcome to AITSM show. Today, we'd be talking about how to make good automation even better. We had joined with Mr. Girish Gadage From ATOS, who is working as a tooling architecture head and a rich experience in a lot of areas like pre-sales, tooling, architectures, many other things being uncovered during the interview.
Thanks, Girish, for have you in our show.
Girish Gadage: Thank you, Anand. Thank you for having me and hosting me on this show. Definitely be glad to share my experience with the audience and with the AITSM community.
Anand Tambey: Right. So in our industries, actually, whenever we are going for automation, so there is a bit of automation in one area bit of automation in another area. And likewise actually starts with some zeal and all, but it fades up when after six months or a few months, and then people already think what to do, but they are not able to actually improvise their automation and make some value out of it. So what are your thoughts and point of view about this?
Girish Gadage: So actually, I would start by stating what you exactly said. Automation is considered as a must have for all organizations. It needs to be driven from the top. Every leader encourages his or her team members in order to do more and more automation in their space. But what when we say, do more and more automation. It should be towards a certain core. The goal could be to Fasten the process. And eliminate manual steps, reduce the number of people required to support an operation or a support ticket, or to be able to introduce more automated way of doing things with this certain other data stores, instead of looking it up manually and to avoid what we call as excel work.
We are masters in Excel. We can do VLOOKUP; can do pivots and try to gather a lot of information that can also be done that can also be automated. And you could just get data, the dashboards, and then you could spend your time looking at the data and get the inference.
So that's, that's the idea of where automation should be done.
Anand Tambey: Yes. Yes. I remember the days when actually to compile a report itself; would have taken almost one or two days.
Girish Gadage: And that reminds me of my earlier career days when I was running performance testing in a product company, and we were getting data points . used to spend about 15 hours collating the data points and generating a report.
So we embarked on this automation process where we did not debate on what technologies should we use. We use the technologies on hand, Pearl scripting and Microsoft VBA programming. And we were able to automate the generation of the report down to 15 minutes. So you can see 15 hours to 15 minutes.
That is the time that we saved, which we could spend on more testing and more use cases. Now, if that is the goal that you will keep for your automation. Then you, and of course, your automation is, successful. What, what I've seen is because somebody is pushing you to do more and more automation, you try to do automation without having a goal in mind.
So you try to automate something without thinking on how this automation will help me. And even without thinking end-to-end process. So if I have a particular kind of working, working in a particular team, which has an handshake with other team in order to provide the whole service, typically each of the teams automates their stuff or their work.
Maybe sometimes they use different technologies, and then they stood on tie-ups. So ultimately you end up,
add value to what you have done. So if you look at it from the sake of having automated things, Yes, I have automated things. Is it going to help you? No, it's not going to help you. Then there is no use of that automation.
Anand Tambey: Right? True. True. So what are your experiences with automation, which can be kind of tips? If there is any tips to actually how to set a better goal for automation or likewise.
Girish Gadage: So, what I've seen is, during my career is, we tend to focus more on what technology to use for innovation. Should we use Ansible? Should we use Bash? we use TCO? Should we use shell scripting? But we do not focus more on the business aspect of what is automation; is automation support strategy?
Right. I think the best way of doing automation is to first define a business objective. What is it that you want to achieve from automation? Then you decide, then you look at technologies on hand. Don't go out of the way, just because some technology is the leading technology in the market
but if you have nobody who knows Python, then doing the automation in Python does not make sense. So there's nobody to support it. So don't worry on the technology to be used ; on the business problem. What is the business problem you wish to solve?
Anand Tambey: Yeah. Right. So we have to focus on the business objective, and then we need to actually depend on. Core capabilities and existing capabilities to move it forward. Right.
Girish Gadage: Correct, correct!.
So, for example, I'll give you an example, which I started my career way back in 2002. We The company was a product company and, that point of time was just starting in India. So we are getting thousands of resumes in our inbox and sorting those resumes out; out duplicates was taking a lot of time. So there was a need to maybe automate that or do it quicker. Now at that point of time, since we had the expertise using VBA, which is an offering given by Microsoft office products, we just chose that technology.
At that point of time, if you remember, VB scripting was more in woke; knew VB scripting. So we used that technology in order to automate it, and save a lot of hours and effort to make things simpler. So that's what I call about. Use what you have and solve the business problem.
Don't worry about. Should I use C++, should I use Python, should I use R programming and the buzz once?
Anand Tambey: All right. Yeah, that's really useful there. And how about if we are going for, say, automation and if automation. Not providing the value. Maybe we have set up our business objectives correctly. We have gone with the tools what we have, it is not on track.
Like what could be the reason about it or what, how we can actually make it and improve it. Sure.
Girish Gadage: So in any automation in the industry. What I talked about earlier was very specific case of automation, where it was only one team that was involved in getting things automated. If you look at your current industry and the way we are supporting our customers.
There are multiple teams involved here. For example, if you support, let's say, cloud technology, public cloud technology, or a private cloud technology, you would have somebody who's providing you the tools to support it like monitoring tools or ITSM tools. You would have somebody who's managing the OS. So OS permissions or OS installation, and then somebody who's managing the network and the network ports, the network ACL, the access control.
. Now, if all of these things don't work together, then your automation will not achieve it. So what I've seen in my experiences that the reason that is not everybody is bought into the idea. Not everybody knows the whole bigger picture of what they're trying to do. For example, if I had the automation team, I wouldn't say, okay.
Network team, open the port; network team is not really sure on why they should open the port? So they will go through the change process of asking 10 or 15 different questions trying to resist open a port because it can introduce a security issue in their eyes. And then there is a lot of back, and forth that happens.
And then you miss your goal of getting things automated. Maybe you're trying to save on time. So this course for two weeks, three weeks, and then before you realize it, you're at end of the quarter in your schedule.
Anand Tambey: Yeah. True, True! So to, in order to actually making. Automation better is there any roll of like what are the type of processes which we are automating or like, say whenever, when we actually come to the ITSM area IT service management area, there are a lot of these requests say getting the software installed, there are, there are service tickets raised, and there are many issues.
So what kind of processes do you think, which are most easily to automate or which could be the best to be automated there?
Girish Gadage: I'll tend answer in a different way, different way. We'll answer. But I believe is when you start embark on a journey of automation, you should aim at the low hanging fruits, which has the less amount of stakeholders, because then you can get things done faster.
You can prove the value, you can achieve a goal, and that then shines up and then there'd be other stakeholders interested. Other people interested to extend this to other areas. So, for example, like you talked about rebooting the server or rebooting a server is while it seems that it's a simple task. It is not really simple task because rebooting the server may have production application installed on it; you reboot the server, the production applications can get impact.
So you have to go through all the process of you're taking approvals taking downtime because you reboot a server. . And then this takes a lot of time, and this workflow or this automation that you do will involve a lot of stakeholders and may not achieve the goal that you want to achieve as compared to, as . for example, if you have a, let's say you have an apache server, sometimes your Apache logs get freed up, or the cache is corrupt, so you need to clean the cache and restart the apache web server. That's something that can be easily done because it involves least amount of stakeholders into it.
Anand Tambey: Right? Right. And once these early wins or maybe - fruits are automated, probably those are the kind of, we can actually promote to other stakeholders and getting interested in it.
Girish Gadage: So once you achieve these, it's also really important in automation to be able to demonstrate with figures in terms of time saved or FTE saved or in terms of dollars saved by automation, because those are the figure.
Those are the statistics that, the business and industry, if you say I'm doing an automation using Ansible library, I've written so many loops, I've touched so many servers, , that's something that business is not interested. The business is interested in what is in it for me(WIIIFM), how many, what is the saving that it's got me in terms of or money.
Anand Tambey: Right, right, exactly. So, the technology may be like maybe high-end or something like even the lower end, the thing which matter most is like what is business getting out of it. So in automation. when we go along with some, say business processes, So some business processes are too complex, and many a times it has many stops, many stakeholders. So when we are going towards that kind of process automation, what are the things you will look for and make automation success for that?
Girish Gadage: Okay. So when there are complex automations, it is similar to doing a complex task.
You have to break down those into manageable pieces. So the first thing I would like is, do you know the process end-to-end? Have you drawn the process end-to-end on a whiteboard or on a piece of paper on how the process works? What are the facts? What are the approvals mechanisms in it?
Where is it a deadlock, or where does it takes time to pass through. So that you can also set up maybe you're new to automation. You can have the right amount of expectations from the automation on what other steps do we take in? Where will it take time? For example, if you have an approval process to execute or reboot server or install something, and if the approver does not approve it for two days, but you're saying my automation will do it in two minutes, then your goals and the actual timing it takes does not match, and approver is on leave you have a mechanism of setting up an alternate approver is there and an alternate approach would be sought in order to avoid the waiting time.
Anand Tambey: Correct. Correct. So, yeah, it's a really interesting point here. Like you just not make one flow as automation, but thinking of alternate flows and alternate stops when a process will actually break down or because of some dependencies on other network or other resources and all, what is the alternative path there is.
And probably there could be an automation or maybe. It would be handed over to some human, like somebody sometimes in in conversational IT in conversational RPA. What we do is sometimes it could be handled totally by chatbots, but sometimes if it is bit complex or bit not understandable, then it will move to a person who can actually get the more information and resolve it.
Girish Gadage: One more anecdote that one more point that I would like to say in the automation, the customers that you are going to work with or the stakeholders you're going to work with, the tools that they have put in place are their babies. So, if you if you project your automation, as they will replace those tools, then you will run into the stake opposition.
What needs to happen is how the automation will work hand in hand with the existing tools and help them to be an efficient system. If you project your automation in that way, then you will have a better chance of success.
Anand Tambey: All right, right. So existing ecosystem of the applications so should be supported by any tool or any automation package.
Otherwise, it will be pretty difficult or even users to actually get through the new system, and they may not be ready for abandoning those. Yes. Right. So yeah, that, that definitely that's true. And very valuable. One more thing I would like to ask about is in, say, robotics process automation, and all, we actually go for some, say, Excel automation or backend API automation or back office automation.
However, would that time has come to actually when we should introduce high-end kind of automation like maybe which interacts with users to actually do the automation rather than maybe like chatbots for suppose. I have to raise a request for server install. I will just open my chat bot and do okay a new server is needed. And then they will give up the form. I will submit something, some information, and that server would be ready on VM. Are these latest technologies and is their time coming up?
Girish Gadage: I would say yes to that. Automation is not something that will always be in the backend. Today automation has to work hand in hand with the humans because, at some point of time, you would have to take in human input in order to execute the automation; do you trigger it? If you look at it more and more, integrate your automation with a modern model of the human on let's say, what kind of interface should I present to the person based on what is his mood today?
What his schedule been today? present a better interface. which may suits him or calms him down; he will be able to make better decision for the questions that you're asking as compared to whether he is in agitated state of mind. So those kinds of how do you get those inputs and then bake that into your automation are becoming more and more important in order to succeed.
Anand Tambey: And sometimes, actually. Whenever these kind ' of new technologies like artificial intelligence and chatbots are introduced, we may have one additional capability there, like understanding users, like, and their past history of requests, where they are coming from, whether they have logged on to a particular system or not, they have some information pre-information about the users.
So they may be better say Okay. This is your next step. Say when I say I am requesting a meeting, so maybe meeting will be scheduled between you and me. Okay. But it will open up. Do you want to schedule a meeting using zoom?
Girish Gadage: I'll go little further let's say you take your example, setting up a meeting. Now the automation can look at your calendar and find that common space, and setup meeting at that space. And that's its default, let's say default behavior, but it can also tell you these are other time slots where you could have the meeting.
So that would give the user, you would give the UI, which is automated.
One of the organization that I was working, there was a default mandate to have one-on-one with your employees. So they set up an automation to schedule one-on-ones, and their automation just used to randomly scheduled it on our calendar without looking whether you're at a different meeting? Whether you're on holidays? Is it Sunday? Is the person available or not. And that is causing a lot of frustration. And what happens is that it's the automation that gets blamed that the automation is not working. It's not that automation that is not working; haven't really thought about it into it, on how this should, how this process should work.
Anand Tambey: True. True, true. Oh yeah. The the framing of requirements and framing of business objectives.
Those are the kind of critical things to actually go for automation. Can you summarize, actually in a few points for our audience that how to actually make good automation better? And even best automation.
Girish Gadage: The first point is you should be aware of your business outcome.
What is the outcome that you want from automation? In order to achieve that outcome, you need to know what is the business process that you are going to automate. And when I talk about business processes end-to-end business processes, third is know what are the points where in you'll need human intervention and the stakeholders involved in the different steps that you'll take to automate.
And then you start automating, you break down the problem and start automating it in pieces. But keep it flexible so that you can join the pieces later. So if you are going to distribute it with different teams and you have inflexible automation, then you won't be able to join it anymore. And your endeavor fails. So keep it flexible so that even if it's different teams working on different pieces, you can join them up together.
And the last point I would say. is always test continuously. Do not wait for the for the whole big picture to be completed and then run the test, and you'll see 110 bugs there, and you will lose heart. So always keep on whenever you develop these small processes, testing them continuously, and then when you join them up or integrate with other processes. Keep testing them so that you are sure of what's happening and how the whole process works.
So that I would say these five steps would actually help you to define and get more value from the automation and also present it back to the stakeholders.
Anand Tambey: Great. Great, amazing, and useful insights we have today with you. And definitely, it would help our audience, and while we are concluding our discussion, what is the one question you think I would have asked additionally to whatever had been asked?
Girish Gadage: I think the one question that's has to be asked is what has been the highlight in your journey, what has been the highlight of your automation, or or what what's it that has made you successful in automation area? I would answer that. Yeah. Let me answer that question that I asked to myself from you. the highlight is when you keep on automating business processes with an outcome in mind.
You get called upon, or you get called as an automation guru because the automation that you make has helped the business. So you get called upon by the C-levels or the CX levels. Whenever there is any kind of automation to be done to decide over it or to build your thought process to how things should be automated.
I think that is I had an experience in my previous company. When they wanted to automate something at a CEO level, and then the CEO gave me a call, why don't you join in and sit there and guide us on how to do this.
Great, thanks. Thanks a lot. Girish proud to have you on our show. There are many topics there where we can actually get benefit of your experience. So we will keep coming back to you.
And if you have more time, we can book more time with you also.
I'll be happy to share my experience for AITSM Community.
Anand Tambey: Right. Thank you. Thank you very much. So audience, if you love this conversation, so please share it generously and give us feedback.
What are the questions to be asked to our IT pros and we will keep coming with more and more interesting topics. Watch this space for more. Thanks. Bye-Bye!