MATLAB: Usage Analysis to Drive Cost Savings

[0:05] Nix: Good morning, good afternoon, and good evening. We welcome you to today’s webinar, MATLAB usage analysis to drive cost savings. My name is Nix and I will be your host. We are so excited to share about MATLAB, which is widely adopted by leading tech firms, manufacturing giants, and engineering companies for complex mathematical modeling, data analytics, image processing, and scientific exploration, among other applications. Since adoption is widespread, even though the cost per license is not high, many companies have a significant investment in their MathWorks licensing. And so we’re glad that you could join our solution architect Linda Cole for an in-depth session on leveraging license usage analysis to generate comprehensive reports and visualization of your MATLAB utilization.

[0:58] Before we start the webinar, let me inform you that we welcome questions. You can type them in the Q&A tab which you can see at the top of your screen. We will try to answer them during the Q&A or get back to you via email. So our presenter has been an industry speaker since 2014, based in Houston. She is a high energy senior level consultant with more than 35 years of diversified experience in the tech industry. She has been consulting clients for the past 10 years here at Open iT on the benefits of software usage metering and how to utilize the various data points to optimize licensing. She is a proven leader with a unique blend of responsibilities in small and large companies with a passion for building successful businesses. Currently she is helping clients solve their business issues around software licensing. Ladies and gentlemen, please welcome Linda Cole.

[1:53] Linda: Thank you Nix, I appreciate that intro, and hello everybody and welcome again to our webinar. I want to first start out by thanking you for being able to attend today and let you know the reason we needed to reschedule is I ended up in the hospital on the date of our original date. I am fine now and I appreciate everybody being flexible and being able to reschedule for today. So as we mentioned we’re going to talk about our focus on MathWorks today. And we’re just going to spend a few minutes on the licensing types because there’s various different ways that you can license the products. And then we’ll talk about some optimization strategies and I’ll go through several different examples to kind of help you visualize what we’re talking about there. And then as Nix said we’ll have a Q&A session.

[2:43] So let’s just jump in. So with MathWorks they have various different licenses. They have what they refer to as standard, or I refer to it as commercial, they have academic and home. They’ve done a really great job of making their licensing available so that when you graduate from college you already have a lot of experience. But today we’re just going to be focused on the standard type of licensing. They have different license terms, right, you can have perpetual, you can have annual, and they have something called pro-rated, but again we’re going to focus on the more common which is the perpetual or the annual licenses today. And they have different entitlement types, so they have what’s referred to as a network concurrent user, that’s where you have a license server and your users check in and check out a license. They have network named user which still runs off of a license manager. They have individual named user or host bound. So we’ll talk about those various different license types and when you should use what type to get the most optimized position.

[4:02] Now when we look at the difference between perpetual and annual, the annual is 40% of the perpetual rate. So if you do the math, the maintenance on perpetual is 20% annually so it’s an exact three-year return. So if you’re going to keep the license more than three years it ends up being less expensive to do perpetual, you know, if you’re three plus years, otherwise you want to do annual. And what’s wonderful is you can do a combination thereof. So let’s say you have a need, an ongoing need, but you have a new project that comes in, you can just put that project on an annual and have your other ones on perpetual. So I think they’re one of the easier vendors to work with, but there are a lot of different ways that you can license that and so you can do any combination thereof.

[4:55] Now when we look at the networked concurrent, that’s where as I mentioned that’s simultaneous concurrent users, so you’re checking a license in and out from a license manager that runs on a license server and you can have it on up to two machines. But the entitlement, again, when you hit that max, let’s say you own 100 licenses, the 101st person that tries to check that out of course will receive a denial. The network named users, this is again, the license is tied to an individual person but you can only run it on one machine and you designate that via the option file. So it’s much, in my opinion, it’s much much easier to have a network named user than an individual named user. Now on an individual named user you can install that on up to four machines but you can only use it on two simultaneously. But again if you want to reallocate that you can reassign that license up to four times per year. You have to uninstall, you have to reinstall, you know, whereas if you have a network named user you can just utilize the option file to move people in and out of that named user license. And then host bound is locked to the machine. Now there are some instances where that makes sense. So obviously every organization is different.

[6:30] But I do want to let you know I’ve seen a customer where they’re actually evaluating, they have a huge MathWorks portfolio and they’re actually evaluating their license position every 15 days, so twice a month. And they’re utilizing both network concurrent as well as network named user, and so they’re moving those assignments in and out to get really the ultimate optimization position. Most people don’t have the manpower, the bandwidth to be able to do it that frequently, but you can come up with any combination thereof.

[7:11] So when we talk about an effective license position or optimization strategies, I don’t care what vendor we’re talking about, MathWorks is going to fall in just with all the others. Absolutely first and foremost we want to identify those applications, those toolboxes, that have zero usage. Now a lot of times what I see is there’s so many different licenses, different vendors, different things coming up, that a lot of companies, especially historically, they would kind of rubber stamp, I’m worried about compliance so let’s rubber stamp our maintenance and support, or let’s just renew what we have, it’s working. But really, why would you want to keep paying for things that you’re not using? So the recommendation is again, with any manufacturer, publisher, vendor, you want to identify those applications that have zero usage.

[8:11] The next thing with MathWorks is because we have those named users, and a named user license versus a concurrent license is about 4X, so it’s much less expensive to buy a named user. So I see customers do this a lot where they have a tendency to use the individual license or the named user, not necessarily the networked one. Again I think that’s easier but they have mainly named users, so that’s not necessarily so. Yes the individual license is cheaper but it’s not necessarily the most optimized. So you really want to look at the usage of those individual users and identify what I’m going to call a power user. So I’ve put the threshold here at 128 hours per month, that’s about 75% of your work day. You can set that wherever you’d like, you can move it up and down depending upon your organization. And what you do is we look and see what those usage patterns are and it’s usually pretty easy to pick where that demarcation should be. And so in my opinion it’s customer specific, we want to look at that usage across the board and like I said there’s usually a jump, so it’s pretty obvious where that should be. So we want to identify the power users and then those that aren’t power users we’re going to say those are occasional users. So those power users it’s more cost effective to put them on, I’m going to say, network named user, whereas your less occasional users you want to use the concurrent.

[10:00] The other thing, and I’ve got some examples of this, is if you have multiple servers that are serving out the licenses you want to be able to simulate combining those. You want to look at your whole portfolio holistically. Even if you have regional requirements that require you to have separate servers, etc., you still want to look and see if I had a single server, if I had a single global server, what would that look like. And even if you keep them regionally it will help you load balance when you go into that negotiation, it’ll help you load balance across those various servers.

[10:41] So once we do those things there’s some additional information that really plays into that, and I’ll again show you some examples today. But you want to be able to, the very most common problem we see at new customers coming on board is what we refer to as long checkouts or license campers. So that’s where when you’re using a concurrent license you check out a license and you aren’t using it. You come in in the morning, you open all your different applications, but you’re typically only using one at a time. So you want to be able to identify it, number one, but also alert the end user. So if they’re leaving the licensing open overnight or on the weekends, etc., you want to be able to alert them so that it’ll drive that behavior, it’ll drive that usage down. You could typically, if you don’t have any metering tool, when you start metering you’ll see it and you’ll say oh these people need to be a networked named user or a named user. But once you get them to understand and you teach them and educate them on the cost of leaving these licenses open, then you’ll see that hey, they were just leaving the license open, they weren’t using it.

[12:02] We’re not going to talk about it today, this is just, today we’re looking at the opening and closing of an application or the check-in and check-out. But be aware that you have the ability to actually go to that next level of metering. In our situation we call it LicenseAnalyzer™ level two, where you can see regardless of when they check in and check out a license, whether it’s a standalone license or a network license, because the same thing happens on a standalone individual user, were you actively using the license. So then you can really hone in. But typically we recommend that you start from a strategy perspective, you start with that what we refer to as level one, that check-in and check-out or open and close.

[12:49] So a lot of folks think, oh I have a named user, an individual named user, it could be a networked or on their local workstation, I don’t care, I don’t have to worry about it because it’s a named user. But that’s not necessarily the most effective license position. So once you get their usage information you optimize to that, and then in the next round we go up to the next level, are they actively using it? Does that make sense? So we’re really tweaking, we want to have the right license for the right person at the right time. If you’ve heard me on any webinar I always say the exact same thing. We want to teach our end user community that they can trust that they’ll have a license available when they need it, and again at the right time.

[13:38] The other thing is the same way we talked about, you have the ability to combine perpetual and subscription depending upon your individual needs, you also have the same ability in toolboxes. Most customers report that toolboxes are their biggest challenge. They understand what their usage is on MATLAB or Simulink or the various different components, but they don’t have visibility on their toolboxes, and this really drives up the cost of their whole portfolio if you will. So you can combine, maybe you have users that should be a named user for MATLAB or Simulink, but when they go to the toolbox usage they only need a concurrent, they can use a shared license because their usage is significantly less on that toolbox.

[14:34] So let’s look at a couple of examples. In this first example we have a lot of different license servers and if you don’t mind let me jump over, it’s easier to show it to you live. So we have, and this is all anonymized, so we can drill down into these and see the actual usage. Here we have one that has toolboxes as well as MATLAB, but again, being able to drill down and look at what this usage is. And this is just an overview on this page. What I really go to is, I want to go to my zero usage. Remember that we said that’s where we’re going to start. So I can go over here to this tab and here we have the various different toolboxes and applications that we own licenses but we’ve not used any at all. And so if you use the cost component, so not just the usage component in your tool or solution, you can actually calculate what the savings is. So in this particular example it’s 177% is the zero usage here. And if we look at the portfolio, the total value of this contract is only $648,000. But then if we go to look at what we refer to as the license efficiency, and so we look at various different components, one is what we refer to as business as usual. So that’s where, if you own 10 and you use 10 then you don’t need to reduce it, if you own 10 and you use five you’re going to reduce it to five. So we call that a business as usual reduction. And so here you can, that’s easy, max available, max in use, so you can easily calculate what that is and again calculate what the savings are. So here we have a $259,000 savings, and that includes the zero usage by the way. But that’s now 40% of the portfolio, and that’s just reducing, so stop paying for what you’re not using.

[16:52] And you can calculate other intervals, so we usually provide a 99 and 95, so those savings just go up as you go down. And I’ll show you some more examples of that as we look at some of these other examples. So let me flip through because I showed you that live. So here’s another example. This is a much larger portfolio. This is where we have a lot of license servers, a lot of applications, a lot of toolboxes, it’s more than I can even put on two screens. But I did split this up. But again we’re calculating, we’re looking at that annual cost. We do have, I’ve got this sorted by elapsed time, MATLAB’s our biggest usage of course. But what does that look like from a business as usual perspective? So if we go all the way to the end of that, so this portfolio value is $2.3 million, as I said it’s a much much larger, more complex portfolio. But again all we did, simple exercise, we did business as usual, meaning let’s reduce to the max that we ever used during our previous, you can look at one year, you can look at two years, it depends if you’re looking at perpetual versus subscription. But I recommend you look at those different intervals to see what makes sense for your organization. But this is a 41% savings, $965,000 out of $2.3 million, 41%. And I have to tell you this is not unusual, we see 30, 40, 50% on a very regular basis for our new customers when they start doing the analytics and looking at the usage of their various different portfolio.

[18:48] Now this particular portfolio has so many different things, you can see there’s duplications. So I resorted it just to kind of give you an idea alphabetically. So here we can see there’s multiple MATLABs, there’s multiple coders, there’s multiple toolboxes. So as I mentioned before we want to look at those holistically. But again, even if you’re forced into having multiple servers, you want to be able to look at those and combine those. So this example I’m going to give you shows you multiple MATLABs, so there’s a lot of different licensing, and this one also has different servers. And we’re just going to look at the MATLAB portion of it. So here on a $3.2 million value, just doing the business as usual is a 50% savings. Again that might seem extreme but I see this on a very regular basis.

[19:57] And let’s just drill into this particular server and we’ll do that same kind of exercise but show you some other data points as well. So if we look at that license efficiency that we talked about, this is that curve, that long tail, we own 70 licenses but we’ve only ever used 25 during that time frame. And 99% of the time we use 22 and 95 is 18. But again we give you so many data points that you can make a more informed, well-rounded view to make a more informed decision. We can see that that 25th license was only used a quarter of an hour. So that gives you again a data point to make you more comfortable, because most people would look at this and say wow, that’s a huge reduction, I don’t feel comfortable with that, right. Even if they were looking at a two or three-year period of time, they don’t feel comfortable. Well when we add this additional information it gives you again a better feel for what’s going on. If we’re looking at the 99, maybe I don’t, maybe I want to license somewhere in between. But having the exact information, how long was that license used, I mean even if we go to the 22nd it’s only an hour, right. So over whatever period of time you’re looking at you can make that more informed decision.

[21:33] But not only does that help in terms of giving you a comfort level to be able to on your first pass do a business as usual reduction, but if we look at the actual users and what is their usage and drill all the way down to the day, you can also go all the way to the hour. But we can go all the way down to the day and then we can identify people that, remember we talked about that long checkout or license camping. So here I’ve identified several where if they were using a license more than 12 hours. Now obviously you will have some users in your organization that may use a license more than 12 hours, but you can see various different patterns going on. Down here we can see this user 1748, they just never checked it in. Maybe they’re running a simulation, maybe not. But it tells you, it gives us an idea of one, how tight am I licensed. I’d feel very comfortable reducing my license from 70 to 25 in this scenario, and then after that, you know, the next renewal or the next true up, I would want to dig in even further to look at this information.

[22:54] Now of course you don’t want to wait, you want to drive behavior. So I would want to set up alerting to help drive this down, hey, you’ve had this checked out for a long time, if you’re not using it please check it back in. Doing kind of a marketing or communications to your end users, I would guarantee you most of your users, unless you have this in your onboarding and training, etc., most users don’t understand the cost implications of coming in and opening a large number of licenses just in case they need them that day, right, they just don’t understand. So helping drive that so when we get to the next true up or renewal we can optimize even further.

[23:39] So we have customers that have been customers 20 years, and so I know for prospects that I’m talking to they’re like, okay so I can maybe see the value year one, but how do you get value after that? And the reason you get value is because your usage patterns change, your organization’s changing, it’s growing, it’s moving, it’s changing, you’ve got mergers and acquisitions, all kinds of things are happening. But the other thing that’s really come across the last few years is the vendors are changing, right. So not necessarily, I mean MathWorks has had the same type of licensing and the flexibility, but when you look at other vendors they’re forcing you to change from concurrent to named user, you don’t have a choice anymore. So again the value of being able to meter these applications and look at this data is key in terms of understanding what you need in your organization.

[24:46] So we talked about combining servers. We have the ability to simulate this, so if you have multiple individual servers you can simulate it and we actually calculate the global concurrency. So again you can look at that. I just wanted to give you an example. We talked about various different strategies of what’s going on, you want to be able to do that, but you don’t want to have to spend a lot of time doing it. So I know we went through that fairly quickly. Hopefully those examples helped bring those suggestions into place. But from my next steps for you, you want to make sure you’ve got your usage metering set up in your organization. You want to identify the zero usage. You want to analyze the usage and see who needs what type of license. Do I have people on a named user that could use concurrent, or vice versa? Am I using up all of my concurrent licenses from power users who really should be on a named user? Identify, do you have people that, we didn’t talk about license hogs, that’s where you have multiple licenses checked out simultaneously, but do I have people doing that where they’re trying to work on multiple things at one time but they’re not truly actively working on it? Do I have people doing camping on that license? And setting up those alerts to drive that optimization. So my goal today was to give you a high-level view of what’s going on and how you can analyze your MathWorks licensing. But that brings us, we only have a few minutes left, so that brings us to our Q&A session.

[26:30] Nix: So why pay for what you’re not using. Thank you so much for your insightful presentation Linda. As you’ve mentioned we can now start the question and answer session. Remember to look for the Q&A tab at the top of your screen and just drop your questions for Linda over there.

[27:11] Okay so I see, yeah go ahead. All right so here from Shelly, she is asking how do we set up metering.

[27:18] Linda: So great question and thank you. I’m going to talk from the perspective of Open iT, we have a product called LicenseAnalyzer™ and it’s a great question because even if you have a tool you may not be getting all the data points, etc. But in order to effectively set up metering for the various different things we talked about, you need to have a data collector that sits on your license server, that’s where your license managers reside. You put a data collector there, and then if you have individual license or host bound licenses you put a data collector on that workstation or wherever the work’s being performed. So if you have a terminal server, application server, right, we’re going to stick it there. So that’s the way to think about it, wherever that work is being performed. And those data collectors report back into a core server. We actually provide the data in various different formats, it’s pre-aggregated, you can report in Microsoft Reporting Services, etc. But that’s how we handle the metering in our tool, in our solution. So I hope that answers your question. But I will say I think we have a slide in a second, we do have, you can use the QR code in a second, or you can go to our website, we do offer free 30-minute consultations with our consultants. So I would encourage anybody, if you don’t feel comfortable asking questions here, please feel free to go and set up a meeting with one of our consultants and we’ll be happy to talk about your specifics in your organization.

[29:13] Nix: Hopefully that is helpful for you Shelly. We do have here some questions beforehand. In my organization we have all of those types of licenses, what do you recommend to optimize?

[29:28] Linda: So we kind of touched on that before. There are situations and organizations and cultures, etc., where you would have to have that kind of complexity. You’d have to have license servers with concurrent, with network named users. But maybe you have some legacy named users that are individual versus host. Now I will say that in most regions across the globe, MathWorks will allow you to convert an individual named user to a networked named user, so you do have that capability, you should just talk to your account manager. So again that’s my favorite, I kept saying that it’s just easier because you can move people in and out with the option files instead of uninstall and reinstall and moving and keeping up with it. But for the individual, I think this came when we rescheduled the original one, you know just contact us so we can talk about your specific needs. That’s what I would suggest. And I think we’re out of time so if anyone has any other questions please feel free to send them in and we will get back with you on that immediately.

[30:52] Nix: Thank you so much Linda, that was very enriching. We also want to thank our attendees for joining us today. Like Linda said, we can answer your questions via email. A quick reminder that this webinar is recorded and the recording will be sent to your registered corporate email address. We’ll also upload it on our webinars on demand page on our website. You can read our new blog related to this webinar, augment MATLAB license usage reporting for strategic advantage. Also check our socials, don’t forget to follow and subscribe, Open iT, Inc. Send us an email at getstarted@openit.com, that’s getstarted@openit.com, to get started with your software optimization journey today. Once again this is Nix, your host for today. Thank you and stay safe. Thank you everybody, have a great day.