Speaker 0 00:00:09 This is the workflow show stories about media production technologies, discussions about development, deployment, and maintenance of secure media asset management solutions. And one of the tools and your workflow therapy toolbox I'm Jason Whetstone, senior workflow, engineer, and developer for Chesapeake systems. Today we'll continue a discussion that Ben and I had with Clinton foundation, chief technology, officer Eric White. We decided to break this discussion into a few parts. So check out the first part of our discussion. If you haven't already, this part of the discussion focuses on some differences between traditional on-premise and cloud infrastructures platforms and applications. We'll define some of those AAS acronyms like SAS and IAS. And we'll talk about strategies for building or even transitioning services to cloud infrastructures, such as virtualization of services and storage, a reminder at workflow show on Twitter and LinkedIn, and please subscribe to the workflow show. So, you know, when to get some more workflow therapy, now let's get back to our discussion with Eric.
Speaker 0 00:01:13 Let's change gears a little bit and start looking at some of the, some of the actual technology. Let's, let's talk about some of the different system types that we see and, you know, especially today, the reason I think this is so interesting and important to talk about is because in the last year, as we all know, our work style has changed quite a bit. And the importance of working remotely working from home has really moved to the forefront. So that often involves some sort of a cloud component. So let's talk about some of the differences between, say an on-premise system versus a cloud-first environment. And, uh, let's, let's start defining some of these AAS acronyms, like I, as pass says, you know, let's talk about those a little bit. So let's start with what, what, what a lot of us are familiar with, which would be an on-prem system, where we have storage servers, uh, maybe a ma'am, maybe a transcoding system, something like that, all within the walls of our building, all operating on our network and let's go from there.
Speaker 0 00:02:13 So on-prem system, we all kind of know why that's challenging and in 2020, 20, 21 and beyond, uh, with the way things are, I, I remember, uh, you did, you did a talk about lift and shift lifting from on-prem to the cloud and some of those challenges, because I think there is a perception sometimes that we can just take these servers that are in our building and virtualize them and put them up in the cloud. And everything's better, right? Well, in terms of lift and shift, lift and shift really implies that you're not going to do a lot of front end work in terms of potentially how to optimize your environment or consolidate services. If you have 40 servers, you're going to do a one-to-one mapping where what those 40 servers might look like running in the cloud just as they are on premise, and whether you're virtual or physical, really doesn't matter that much.
Speaker 0 00:03:10 It matters in how you're going to do the migration. But in terms of the concepts of what lift and shift means, you're basically going to have a one-to-one mapping. Now, when you do that, you're really talking about, I am so infrastructure as a service, you're going to look at, okay, I am X number of servers. How many virtual machines would that map to in the cloud? How much compute do I need? How much memory, how much does, how much network capacity do I need? And you do all those models and they're coming out. All three of the big cloud providers are coming out with good tools to help you figure out things you can run in your environment, uh, Azure, which I just happened to be a little more familiar with no surprise given my background, but actually if you have a VM-ware environment, it will, or, or a hyper V environment. Those are VMware is the hypervisor technology from VMware, IBM and Microsoft of course has hyper V that's their hypervisor basically tools to kind of model what your environment might look like in real time in the cloud without incurring any costs.
Speaker 1 00:04:21 I was just going to say, we should define what a hypervisor is for some of our listeners who might not have run into that before, who might be new to virtualization, who might be familiar with production work and new to the entire world of
Speaker 0 00:04:34 Virtualization and
Speaker 1 00:04:36 What they do on a regular basis with a west of the internet.
Speaker 0 00:04:39 Happy to do that. Hypervisor really just means you're building a virtualization layer on top of a physical piece of hardware. So take a server that has, let's say it's a dual processor, 12 core machine. You effectively have 24 cores. Now it can be hyper threaded. So then you would double the amount of force and those can become your virtual cores that you could then assign. And you can, it doesn't have to be a one-to-one relationship. That's, what's great about virtualization. You can carve them up as much as you want, but basically it's a layer on top of the operating system that exposes resources in a virtual way, that you could then create a virtual instance of a server. So you create an image. It's going to have X number of virtual processors, X number of gigs of Ram six gigs of Ram and X amount of disc. And it will pull that from a subset of what that physical server that it's running on has available. So if you have 128 gigs of Ram and you have seven terabytes of disc space and you have 24 cores that could be carved up potentially even more than in the simplest manner, you would allocate a certain amount of that to each virtual server. And then they run as if they're running on the hardware directly. They don't, they don't know any differently.
Speaker 1 00:06:06 It makes me think of like a dream of a server. It's like the servers dreaming of another server within the server server.
Speaker 0 00:06:14 I like that. That's great. That's a really great description of hypervisor and virtualization. Thank you. Let's go on from there. We were talking about infrastructure as a service and how we can sort of model that on prem before we have to incur any costs. And I'll highlight too much of the Clinton foundation is still on premise for a number of reasons, which I can highlight as we talk about the other types of services that you can obtain in the cloud, that we are in a lot of ways, we still have a lot of things on premise, but we're also very heavy, heavy consumers of SAS. So SAS is software as a service, and that's when you buy a platform that I'll call in simple terms already baked. So some type of document sharing platform, office 365, Dropbox slack, basically a end service product. That's based on the cloud, that's consumable in a monthly or annual way on a per user per month pay.
Speaker 0 00:07:14 As you go basis. Those are all attributes. So what a SAS application would be. And, uh, we just did an assessment last year. It was sort of under the guise of security with all the data breaches and seemingly cloud vendor of the month data breaches. Uh, we of course in advance of doing any type of data governance. We did an assessment where we store data with whom how do they protect our data? And we have upwards of like 11 or 12 cloud vendors, SAS cloud vendors that we use. How does software as a service differ from platform as a service? This is a discussion I've been wanting to have on the show for a while is really like sort of the difference between a platform and an application and why that difference is important. It's an important distinction. There really is a continuum here. So infrastructure as a service is at the core, right?
Speaker 0 00:08:08 If you want to do anything in the cloud, let's say, you're going to be the next slack, right? You have a better idea. You're going to pull together a team of developer, some infrastructure folks, and you're going to build the next version of slack. Well, one place to start is I, as right, you build your own servers, you install your own development software. You kind of build your own from scratch, but it's in the cloud. So that would be a level up from there would be paths, which is platform as a service. So instead of deploying a virtual server, deciding on how much Ram, how much this, how much compute it needs, and then trying to optimize that continually over time, of course, there's tools that help us do that. You have the servers under utilize. This one's utilize more, you know, and rebalance. What platform is a service lets you do is do away with the management of virtual servers, but merely consume services.
Speaker 0 00:09:03 Every platform, Instagram has a database. Slack has a database, right? Where does it store on Slack's messages? Where does it keep its list of users that have access? Where does it keep all the channels? Any of those platforms, even a Dropbox is going to have a database. Yeah. So in a traditional say on-prem environment, of course we have databases on premise. Uh, what, what we would probably have is a server that is hosting that database, that there would be storage that the actual data lives on. And then there would be a service like Postgres or my sequel or SQL server or something like that, that is actually hosting that data and letting the applications and services interact with that data. So what you're saying then is in a platform as a service environment, that database is not being hosted by a server, even a virtual server.
Speaker 0 00:09:53 So in the platform of a service model, you're saying I'm just going to purchase this database as a service. So when you go into Google or Amazon or Microsoft, you can deploy a database without having to first deploy a virtual machine, configure it and install the database software, and literally build it from scratch. You can just press a button, define what the parameters are, that database, what resources you want to allocate to it and just pay for that service. And you're just paying for your usage. Yeah, exactly. Per month per use for that database. So you're not carrying the overhead of a virtual machine, just for the database, similar to our apartment example where you're not carrying the overhead all by yourself, living alone in an apartment, building of all those other apartments, you're just paying for your own. And, and again, this, this gets into the whole cloud versus on-prem discussion in general, really, which is, you know, when, when your application is running on Saturday night and your users are at home on a virtual happy hour call, I guess what hopes yeah.
Speaker 0 00:11:01 Uh, your, your application is just sitting there not doing anything. Those servers are taking up resources, they're taking up electricity and heating up the data center and all that. So we kind of do away with all that, right? We, we, we, we, we don't have to incur that cost because we're only incurring costs for usage. For those, for those services, you could argue they're somewhat more scalable because they're in the broader backend of whatever infrastructure they've allocated to support that service on the backend. And also it's not just database web services. Every web, every internet application has to connect to a server over the web, right? Even if it's technically not a web server that has a web page you would visit, but any internet enabled service is going to have any client slack running on your desktop tech doc, running on your phone, WhatsApp, whatever you want is all going to connect back to some server, running a web web interface.
Speaker 0 00:12:01 And so you can deploy these web interfaces as a service, just like you can a database and again, not having the overhead of a server. So in a perfect world, if you are planning a move to the cloud, you would not deploy VMs. You would literally port your database. If you have an accounting database, you import your database to database as a service, any web applications you would port to a web app as a service, any other code that you're running, you know, natively, whether it's Java or what have you, they have repositories where you can run those things. And so not saying it will work in every instance, because there are times where you need physical control over the outer environment of that database and there's other considerations. But in an ideal world, that would be the most optimal environment would be to just leverage services and not, not move servers.
Speaker 0 00:12:53 Right. And it strikes me that this is a discussion that we often have when we talk about certain platforms and solutions. And when we use the lingo like, well, this application has a really modern architecture or this application, or this solutions architecture is a little dated. A lot of that has to do with, um, traditionally some of these, some of these solutions were developed to run on physical hardware. That's just the way they work. So when we talk about, can we run this on the cloud? Can we deploy this on the cloud? A lot of it goes back to, well, what is the architecture? Is the architecture designed to be run on physical servers? In that case, we may have to go with VMs, right? And then we're really not utilizing the full potential of some of these, what we would say as a modern architecture, where we are using a database in the cloud, we are using web application being hosted by a cloud service.
Speaker 0 00:13:39 So just to sort of speak to that a little bit, when we talk about a modern Vertek detection architecture versus a, maybe a more dated one, that's what we're talking about. Yeah. W Y just to dovetail off of that, you make a good point in that if you're looking at a, a technology solution and you're going to buy a product of a technology product that otherwise was intended to run on premise, it may not necessarily, you know, if it's built from the ground up it as its own client, that's its own database it's meant to be installed on a server. It's not going to necessarily be compatible with that platform of the service architecture. Yeah. And that speaks to the fact that you may see a really cool feature in a platform that you really like, and you may say, well, we want this platform because of the feature set, but we really need to run it in the cloud. And you know how that can be a challenge sometimes because some platforms aren't just, aren't designed to utilize that modern architecture. That doesn't mean it's not going to work. It just means that there could be challenges. Number one. And number two, you may not get the same benefits that you would get if the platform was designed to be run in that cloud first way. That's right. Yeah. So what,
Speaker 1 00:14:47 What that brings to mind for me is that one, it, it might make sense to move to the cloud, even in those instances, if you're trying to avoid things like hammering on the wide area network access from, you know, the enterprises, internet pipes, right. If we're shifting over to a cloud, then we're using the cloud providers pipes, and those pipes are really wide. So that's always a big advantage there, but then it also brings up the, the underlying storage and the question of performance and how to build all of these things to take advantage of some of the platform as a service offerings. And that's where we really get into whether or not something is built for the cloud specifically just about anything can be run in the cloud, but whether or not it's built to use the underlying structure of those cloud services really is a differentiator.
Speaker 0 00:15:39 Another challenge I find that we have with, with some, some of these solutions is, is that the hybrid model where we want to work with a cloud first application, maybe, um, just to sort of throw an example out there. We may want to work with the cloud first application, but our data meaning our high res video files for example, are living on an ass or sand within our walls on our land. And now we want to offer that up to maybe some editors working from home, or maybe we want to share that out to a, uh, another work group or a client or something like that. And the challenges that we're presented with there are, well, how does this cloud application talk to our solution? That's running on prem. We see this a lot now in this modern day and age, because, because of the whole work from home, the push to work from home and work remotely.
Speaker 0 00:16:27 So to sort of carve this example out a little bit further, let's say we have a really great modern architected platform that is running in the cloud. Let's say it's a media asset management platform. We want to use that, that cloud first application, we're really happy with the feature set. It's a really cool piece of piece of technology. Our users are really psyched and excited about it, but all of our high res video files live on a server inside of our organization. We know we can make proxies and put those proxies up in the cloud as well. So that at least, you know, that the users have something to look at. But what about when we need to do some work on those files that are living in that organization? The man has really kind of it's up in the cloud and it's, it's, it's doing the ma'am tasks, but we have something within our organization that needs to touch those files and do the work.
Speaker 0 00:17:11 Sure. That can be a challenge as well, right? Absolutely. Yeah. It would definitely have to architect around some way for them to be able to work locally and then pass information back and forth, whether it's native to the media asset management platform, whether some workflow has to be developed in order to do that lightly, depending on their existing internet bandwidth capabilities within the office, you'd have to look at, you know, a bigger pipe potentially to support the passing of large files back and forth. Some systems that check out, check in capability where you're kind of checking something, kind of download it, you know, kind of work on it and check it back in. And a lot of that has to be addressed. I think from a workflow perspective, another potentiality could be too that you could build some type of remote access capability, uh, whether it's remote desktop, where they're actually logging in to the cloud and working on everything in the cloud where they're not passing these large files back and forth all the time. And then that can help optimize things to a degree as well for some virtual desktop solution that brings
Speaker 1 00:18:25 To mind the idea of a private cloud versus a public cloud, right? We'll talk about the cloud service providers being, you know, Amazon and Google and Microsoft, and, you know, these big tech companies that run these giant mega complex data centers that can really grow. You know, if you start out, oh, I'm an entreprenuer and I've got an idea for a business and suddenly somebody, you know, posts about your cool new thing and it goes viral. And if you're using these modern architectures, they have the ability to scale in the background, right? They might be running in a containerd environment. That means that it can be, uh, can be injected stateless Lee into these containers. And then that you can take advantage of the clouds economy of scale, um, so that it can really balloon up and be able to handle the volume that you might be seeing with interest within your platform.
Speaker 1 00:19:21 However, you might not be able to pay for it. But that's another story when we're talking specifically in the media and entertainment industry, the, the idea of these high resolution media files and the qualities with which we need to work in order to do that fine creative work, that's really where the rub is in how fine of a quality do we need, right? With proxy video, we can download it or stream it, and we can often get pretty decent quality. Like if you showed me this, what we're doing right now, I think I've said it in the show before, you know, just by being able to talk to you guys on zoom, the quality with which these tiny pictures I see you is rather amazing. Um, and 10 years ago that would have been astounding, but is it good enough to, you know, finish a feature film, maybe not.
Speaker 1 00:20:10 Right. Um, so that's where it we'll talk about something like a potentially a private cloud, meaning that we might want to build. What ends up being more like an on-premise infrastructure somewhere else that you get some of the benefits that it organizations are looking for in terms of not wanting to have to maintain and govern that hardware themselves, where you still might get some of the performance, because the underlying storage is fast enough to handle the files. You can run whatever software you need to just because it does mimic a traditional infrastructure in terms of being servers. And maybe if you're lucky you can band with other like-minded individuals who need that same service and kind of start to leverage some of those economies of scale.
Speaker 0 00:20:53 That's right. Actually a big fan of more boutique players when it comes to cloud initiatives, looking at different ways to store things in the cloud. Like I said, big fan of private cloud providers, because they can give you the, again, the convenience, the economies of scale, but you don't have the layers of dealing with a large organization to get certain types of support, particularly when it comes to more, more unique types of configurations like media, asset management, uh, that have their own quirks and idiosyncrasies to them to work with somebody that has knowledge of the whole stack, right? Yeah. As knowledge of what it takes to stand up a private cloud and also the product that they're supporting and they can do both. That's a win-win situation. And so I think there's a lot of value there for sure. That's it for the second part of our discussion with Clinton foundation CTO, Eric White, thanks for listening.
Speaker 0 00:21:58 And the next episode of the workflow show, we'll bring back Dave Helmsley from Adobe and also introduce his colleague Michael Gamble. We'll be discussing a new partner certification program. Adobe has recently brought online called the Adobe certified service partner or AC ESP program. We'll be discussing a new partner certification program. Adobe has recently brought online called the Adobe certified service partner or AC ESP program. Chesa is proud to be one of the first US-based systems integrators to receive this certification. We'll discuss with Dave and Michael. What's so cool about this program and what it means for Adobe and for the media industry. The workflow show is a production of Chesapeake systems and more banana productions. Original music is created and produced by Ben Kilburn. Please subscribe to the workflow show and look us up on Twitter and LinkedIn at workflow show, email, workflow [email protected]
Thanks for listening. I'm Jason Wetstone