10/23/24 - Episode 5 - Demystifying the Cloud

Text Summary

Eric Snyder: Hi everyone and welcome to webinar number five I'm Eric Snyder and I'm the Executive Director of the University of Rochester Wilmot Cancer Institute's Technology and Innovation Group today.  I'm joined by our head of architecture Scott Isaacs. Scott is one of those people that has been all over the place um seen a lot in the technology space in general.  He's the former director of the Cyber Collaboratory down in Louisiana and SMNE for CompTIA and has a pretty vast knowledge in the space in general , so he's going to be talking about the cloud today um this is a presentation he did at HIMSS a couple years ago and for anyone who's been to HIMSS or who has really any almost conferences are the same you know you generally get 20, 30, 40 people in in the audience that this this presentation I think had the most we've ever seen outside of a keynote at it was 100 plus easily with some pretty incredible feedback as well. Lastly before we get started here November's webinar is going to be back to our technology releases with uh the expansive Data Dictionary application and it'll be end up presented by lead developer Emily Strong so we'll have more on that shortly but without further ado here is our head of architecture Scott Isaacs.

Scott Isaacs: Thank you Eric I appreciate the introduction, like Eric said my name is Scott Isaacs, I’m the Head of Architecture and Senior Innovator here at the Technology and Innovation Group and I'm going to be giving an abbreviated version of my Demystifying the Cloud you're just renting somebody else's servers talk from HIMSS.  Today's agenda is going to be, we’re going to start off with just a high-level overview of what uh Cloud infrastructure is and the technology behind it then we'll get into some Cloud utilization strategies.  

So to just go ahead and kick us off what is the cloud? The cloud is an ever-evolving infrastructure of different data centers that are connected software microservices and things of that nature.  Every day uh different Cloud providers from Azure, AWS to Google come out with new different offerings or change their offerings uh just to kind of give you a basics of what the comparison of on premise versus cloud different models are, in the beginning there was three you had infrastructure as a service platform as a service and software as a service and as you can see from this chart,  depending on what the offering is depends on how much of it is managed by yourself or how much of it is managed by your cloud provider um and offloading those different items that are managed by the cloud provider is just less for you to have to do but it's also less control that you have well and since this was how the cloud pretty much started with these three services it's actually expanded over the past decade or so to include things like STAS, Pas, Das so, uh and you can see here with this funnel the lower it is the more control you have and the more but the more you're responsible for and the further up you go the more it's all controlled by the cloud provider and I'm sure you can see some things that you recognize on here uh like uh Office 365 and G Suite those are your software as a service things then you have functions as a service which are just microservices to where uh there's a little trigger that says hey we saw something we need to do something about it,  and then well database is a service which you know we've all use databases so as you can see there's a lot of different things. Then of course storage is a service I for to mention that one uh that's your you know Dropbox and things of that nature.

The thing with this expanded model is that all of these actually existed beforehand under the three models, they've just kind of subbed them out into a uh more Niche products,  and the one of the problems with that with current cloud infrastructure is that over the years a lot of this Evolution has been nomenclature and marketing driven and I say that as in  they will rename services that already exist. So, like a private Cloud, private cloud can be two different things it can be an instance that  you have privately in Azure or AWS, or it could be a cloud that you have on your uh on premise which that's basically just data center virtualization at that point but now it's been nicknamed private cloud and then you have manage private Cloud. Edge is one of the newer one that's come up um and then DAAS which you have to worry about with Daas is it used to be uh database as a service and but now it can be desktop as a service as far as a virtual desktop environment. So, uh and now they are you know just using new names for existing Services which if you don't have Cloud experts on hand it leads to a lot of confusion and organizations because if you're not keeping up daily with these new um changes that are coming out you can get left behind or use the wrong nomenclature and spin up the wrong items,  so it's something you really have to keep up with and like I mentioned before uh just with SQL database instances it used to be platform as a service but now it's database as a service so which you don't need to confuse with desktop so having Cloud expertise under the roof is extremely important in today's Cloud’s postures for deployments. So, and then of course Edge Computing is one of the newer ones uh what used to be known as IOT and is still known as IOT Internet of things is now incorporated into edge computing, so it says it's because so your smart thermostat that used to be iot is now considered Edge because it has its own compute processor inside of it. With a lot of the where Edge used to be just be like an actual server something that would process somewhere in between a sensor and a data center so, once again trying to keep up with the terminology and all the changes that are within the cloud is something that's in of itself is a full-time job.

So, let's go into utilization strategies, with the cloud things are always evolving so, uh Cloud providers are pretty much trying to figure out how to use their own services as they're deploying them to you because there are a lot of instances where the backend changes, security changes like said you saw the nomenclature changes or they will deprecate a service for another service. We recently had it to where the iot buttons that one cloud provider had sold to the hospital had been deprecated and they're trying to figure out how to replace it because they had a lot of micro functions tied into that for operations and it was something that was you know when you have hundreds of these uh buttons and sensors out there and then a cloud provider deprecates them and you're reliant on them, you've got to be able to shift very quickly to get uh at a replacement or compensate for that, so, one and very important thing is never use preview of any production workloads I've seen preview I've seen a preview for a particular product go on for six years and then be dropped by the cloud provider,  it also means it's not stable so there's also a lot of problems with it and it also means it's going to be changing frequently, so,  the way you're using it now might not be the same way they'll be using it in two months. So, preview in an Enterprise sort of solution you never want to use anything like that even though it looks like it might fit your needs.

One thing that gets a lot of people uh caught up in trouble in dealing with cloud is that the terminology is the same for a lot of on- premise a assets, but they do not function the same so you really need to do have experts that uh work in Cloud as opposed to your on premise because you'll have a lot of organizations say we have an IT person that deals our on premise data center this all looks the same they can do both and that's not the case,  having that specialized expertise in somebody that can look at the cloud and know how it functions and works because uh v-net in a data center versus of vet and the cloud even though they're labeled the same and the principles the same they're actually set up differently.  In many Cloud initiatives that I've seen have failed due to not investing properly in the internal talent,  because that internal talent is imperative to understanding the cloud, keeping you out of trouble for like I said a different production workloads and the way they change and also budget overruns I've seen some massive budget overruns because something wasn't scaled properly, you know to  the tunes of tens of thousands of dollars being lost because a uh asset was accidentally set as a static asset where was fully deployed all the time instead of being a burst asset to where it's shut down automatically afterwards.  So, is not only can you not have your Cloud deployment malfunction, you can also run up large costs uh if you're not careful.  When it comes to your governance strategies while stakeholders are important they don't really understand how all of this works as far as the cloud is concerned and by  stakeholders I don't mean just mean your standard IT department and the hierarchy of it,  I also mean as far as your different departments and things of that nature that are going to be utilizing the cloud assets so,  I said while they should have a idea of what's going on and have input your final strategy should lie with your subject matter experts because how fast and how flexible the cloud is, you can't really go to committee every single time and because committees is where Innovation goes to die pretty much because you just get stuck in a loop of let's revisit this,  let's revisit that,  you want to try to make everybody happy and in that sense nobody's happy and then you'll never move forward.  So, you get wind up getting stuck with the sticky middle where everybody wants to compromise on it and then you wind up without a product and the cloud changes too quickly for that and that's actually not the purpose of the cloud is to actually move fast and get things done.

You really do want to make sure that you uh take care of a do a cost benefit analysis on your different workloads because if you  people have been finding and retracting from the cloud Zone because they found the expense was more than they thought it was going to be compared to on-premise assets. If you're doing something you know very specific workloads that are burst because you want it to have access to some high compute power but only for a short duration, that might be good for a cloud, but if you have some equipment that is going to utilized significantly for a month or two, or you know a year actually it might be better for you to purchase the equipment and then have it on-premise, you’ll see significant saving and have more control over it especially in highly secure environment’s. It’s like a hospital and like I said burst computing, especially workloads does make sense or if you have some archive storage it,can be cheaper than enterprise solutions that you have on-premise, but you need to check against how much it would cost internally to purchase it, especially with doing GPU’s and AI in the cloud are fairly expensive if you’re going to be utilizing them frequently, checking on your own workloads and your data center purchases is very expensive.

So, just a recap, you need to make sure you have a dedicated team that stays up to date on your cloud infrastructure and the different changes to it at all times, and the different offerings that are coming out and the different offerings that are coming out because some might be better for your workload than what you're doing currently, so yeah many Cloud assets do have the same naming conventions as your on- premise assets but that doesn't mean they work the same that doesn't mean that the people that take care of your on premise assets would automatically be able to take care of your Clouds. So,  you really you need to honestly assess your internal cloud expertise and staff accordingly to make sure that you have the correct people in place and enough of them to do the work because you don’t want to fall in the yeah we can do that trap because there’s a lot of well-meaning people in enterprise where they want, they can say we can do that and just because you have the capability to do something doesn’t mean you have the ability or the bandwidth to do it, because people their own full time jobs and if they’re trying to take over part of the cloud, they might not have the actual bandwidth to do what they need to do.  So, and then once again make sure you review your burst computing versus purchasing services internally for especially workloads, because you can get into a situation where you’re spending a lot more on cloud services than you are just actually purchasing the server internally and then you would have more availability of it as well.  And finally, do you need to use the cloud, or are you just wanting to? That’s one thing I’ve run into a lot of times to where people assign the fact that the cloud is something more advanced or they think they ought to use it for a particular grant or something of that nature, when actually using the data center makes a lot more sense. They’re just tied up in the notion of wanting to use the cloud or what would be considered: “more advanced services” and so that the short and sweet version of my HIMSS talk.

At this time, we will open it for any questions and on the screen I actually says it a link to our website, a QR code to our website and also my email address if you have any further questions. I’d be happy to answer them for you.