(upbeat music) >> Welcome back to London everybody, this is Dave Vellante with The Cube, the leader in tech coverage, and we're here at AWS. We wanted to cover deeper the public sector activity. We've been covering this segment for quite some time, with the public sector summit in DC, went to Bahrain last year, and we wanted to extend that to London. We're doing a special coverage here with a number of public sector folks. Anjenesh Babu is here, he's a network manager at Oxford GLAM. Thanks very much for coming on The Cube, it's good to see you. >> Thank you.], thanks. >> GLAM, I love it. Gardens, libraries and museums, you even get the A in there, which everybody always leaves out. So tell us about Oxford GLAM. >> So we are part of the heritage collection side of the University. And I'm here representing the gardens and museums. In the divisions we've got world renown collections, which has been held for 400 years or more. It comprises of four different museums and the Oxford University Botanic Gardens and Arboretum. So in total, we're looking at five different divisions, spread across probably sixteen different sites, physical sites. And the main focus of the division is to bring out collections to the world, through digital outreach, engagement and being fun, bringing fun into the whole system. Sustainment is big, because we are basically custodians of our collections and it has to be here almost forever, in a sense. And we can only display about 1% of our collections at any one point and we've got about 8.5 million objects. So as you can imagine, the majority of that is in storage. So one way to bring this out to the wider world is to digitize them, curate them and present them, either online or in another form. So that is what we do. >> In your role as the network manager is to makes sure everything connects and works and stays up? Or maybe describe that a little more. >> So, I'm a systems architect and network manager for gardens and museums, so in my role, my primary focus is to bridge the gap between technical and the non-technical functions, within the department. And I also look after network and infrastructure sites, so there's two parts to the role, one is a BAU business as usual function where we keep the networks all going and keep the lights on, basically. The second part is bringing together designs, it's not just solving technical problems, so if I'm looking at a technical problem I step out and almost zoom out to see, what else are we looking at which could be connected, and solve the problem. For example, we could be looking at a web design solution in one part of the project, but it's not relevant just to that project. If you step out and say, we could do this in another part of the program, and we may be operating in silence and we want to breakdown those, that's part of my role as well. >> Okay, so you're technical but you also speak the language of the organization and business. We put it in quotes because you're not a business per say. Okay, so you're digitizing all these artifacts and then making them available 24/7, is that the idea? What are some of the challenges there? >> So the first challenge is only 3% of objects are actually digitized. So we have 1% on display, 3% is actually digitized, it's a huge effort, it's not just scanning or taking photographs, you've got cataloging, accessions and a whole raft of databases that goes behind. And museums historically have got their own separate database collection which is individually held different collection systems, but as public, you don't care, we don't care, we just need to look at the object. You don't want to see, that belongs to the Ashmolean Museum or the picture does. You just want to see, and see what the characteristics are. For that we are bringing together a layer, which integrates different museums, it sort of reflects what we're doing in out SIT. The museums are culturally diverse institutions and we want to keep them that way, because each has got its history, a kind of personality to it. Under the hood, the foundational architecture, systems remain the same, so we can make them modular, expandable and address the same problems. So that's how we are supporting this and making it more sustainable at the same time. >> So you have huge volume, quality is an issue because people want to see beautiful images. You got all this meta data that you're collecting, you have a classification challenge. So how are you architecting this system and what role does the Cloud play in there? >> So, in the first instance we are looking at a lot of collections were on premises in the past. We are moving as a SaaS solution at the first step. A lot of it requires cleansing of data, almost, this is the state of the images we aren't migrating, we sort of stop here let's cleanse it, create new data streams and then bring it to the Cloud. That's one option we are looking at and that is the most important one. But during all this process in the last three years with the GLAM digital program there's been huge amount of changes. To have a static sort of golden image has been really crucial. And to do that if we are going down rate of on premise and trying to build out, scale out infrastructures, it would have a huge cost. The first thing that I looked at was, explore the Cloud options and I was interested in solutions like Snowball and the Storage Gateway. Straightforward, loads up the data and it's on the Cloud, and then I can fill out the infrastructure as much as I want, because we can all rest easy, the main, day one data is in the Cloud, and it's safe, and we can start working on the rest of it. So it's almost like a transition mechanism where we start working on the data before it goes to the Cloud anyway. And I'm also looking at a Cloud clearing house, because there's a lot of data exchanges that are going to come up in the future, vendor to vendor, vendor to us and us to the public. So it sort of presents itself a kind of junction, who is going to fill the junction? I think the obvious answer is here. >> So Snowball or Gateway, basically you either Snowball or Gateway the assets into the Cloud and you decide which one to use based on the size and the cost associated with doing that, is that right? >> Yes, and convenience. I was saying this the other day at another presentation, it's addictive because it's so simple and straight forward to use, and you just go back and say it's taken me three days to transfer 30 terabytes into a Snowball appliance and on the fourth day, it appears in in my packets, so what are we missing? Nothing. Let's do it again next week. So you got the Snowball for 10 days, bring it in transfer, so it's much more straightforward than transferring it over the network, and you got to keep and eye on things. Not that it's not hard, so for example, the first workloads we transferred over to the file gateway, but there's a particular server which had problems getting things across the network, because of out dated OS on it. So we got the Snowball in and in a matter of three days the data was on the Cloud, so to effect every two weeks up on the Snowball, bring it in two weeks, in three days it goes up back on the Cloud. So there's huge, it doesn't cost us any more to keep it there, so the matter of deletions are no longer there. So just keep it on the Cloud shifting using lifecycle policies, and it's straight forward and simple. That's pretty much it. >> Well you understand physics and the fastest way to get from here to there is a truck sometimes, right? >> Well, literally it is one of the most efficient ways I've seen, and continues to be so. >> Yeah, simple in concept and it works. How much are you able to automate the end-to-end, the process that you're describing? >> At this point we have a few proof of concept of different things that we can automate, but largely because a lot of data is held across bespoke systems, so we've got 30 terabytes spread across sixteen hard disks, that's another use case in offices. We've got 22 terabytes, which I've just described, it's on a single server. We have 20 terabytes on another Windows server, so it's quite disparate, it's quite difficult to find common ground to automate it. As we move forward automation is going to come in, because we are looking at common interface like API Gateways and how they define that, and for that we are doing a lot of work with, we have been inspired a lot by the GDS API designs, and we are just calling this off and it works. That is a road we are looking at, but at the moment we don't have much in the way of automation. >> Can you talk a bit more about sustainability, you've mentioned that a couple of times, double click on that, what's the relevance, how are you achieving sustainability? Maybe you could give some examples. >> So in the past sustainability means that you buy a system and you over provision it, so you're looking for 20 terabytes over three years, lets go 50 terabytes. And something that's supposed to be here for three years gets kept going for five, and when it breaks the money comes in. So that was the kind of very brief way of sustaining things. That clearly wasn't enough, so in a way we are looking for sustainability from a new function say, we don't need to look at long-term service contracts we need to look at robust contracts, and having in place mechanisms to make sure that whatever data goes in, comes out as well. So that was the main driver and plus with the Cloud we are looking at the least model. We've got an annual expenditure set aside and that keeps it, sustainability is a lot about internal financial planning and based on skill sets. With the Cloud skill sets are really straightforward to find and we have engaged with quite a few vendors who are partnering with us, and they work with us to deliver work packages, so in a way even though we are getting there with the skills, in terms of training our team we don't need to worry about complex deployments, because we can outsource that in sprints. >> So you have shipped it from a CAPX to an OPX model, is that right? >> Yes >> So what was that like, I mean, was that life changing, was it exhilarating? >> It was exhilarating, it was phenomenally life changing, because it set up a new direction within the university, because we were the first division to go with the public Cloud and set up a contract. Again thanks to the G-Cloud 9 framework, and a brilliant account management team from AWS. So we shifted from the CAPX model to the OPX model with an understanding that all this would be considered as a leased service. In the past you would buy an asset, it depreciates, it's no longer the case, this is a leased model. The data belongs to us and it's straight forward. >> Amazon continues to innovate and you take advantage of those innovations, prices come down. How about performance in the cloud, what are you seeing there relative to your past experiences? >> I wouldn't say it's any different, perhaps slightly better, because the new SDS got the benefit of super fast bandwidth to the internet, so we've got 20 gigs as a whole and we use about 2 gigs at the moment, we had 10 gig. We had to downgrade it because, we didn't use that much. So from a bandwidth perspective that was the main thing. And a performance perspective what goes in the Cloud you frankly find no different, perhaps if anything they are probably better. >> Talk about security for a moment, how early on in the Cloud people were concerned about security, it seems to have attenuated, but security in the Cloud is different, is it not, and so talk about your security journey and what's your impression and share with our audience what you've learned. >> So we've had similar challenges with security, from security I would say there's two pots, one's the contractual security and one is the technical security. The contractual security, if we had spun up our own separate legal agreement with AWS or any other Cloud vendor, it would have taken us ages, but again we went to the digital marketplace, used the G-Cloud 9 framework and it was no brainer. Within a week we had things turned around, and we were actually the first institution to go live with and account with AWS. That is the taken care of. SDS is a third party security assessment template, which we require all our vendors to sign. As soon as we went through that it far exceeds what the SDS requires, and it's just a tick box exercise. And things like data encryption at rest, in transit it actually makes it more secure than what we are running on premise. So in a way technically it's far more secure than what we could ever have achieved that's on premise, and it's all taken care of, straight forward. >> So you've a small fraction of your artifacts today that are digitized. What's the vision, where do you want to take this? >> We're looking at, I'm speaking on behalf of gardens, this is not me, per say, I'm speaking on behalf of my team, basically we are looking at a huge amount of digitization. The collection should be democratized, that's the whole aspect, bringing it out to the people and perhaps making them curators in some form. We may not be the experts for a massive collection from say North America or the Middle East, there are people who are better than us. So we give them the freedom to make sure they can curate it in a secure, scalable manner and that's where the Cloud comes in. And we backend it using authentication that works with us, logs that works with us and roll-back mechanisms that works with us. So that's were we are looking at in the next few years. >> How would you do this without the Cloud? >> Oh. If you're doing it without the Cloud-- >> Could you do it? >> Yes, but we would be wholly and solely dependent on the University network, the University infrastructure and a single point. So when you're looking at the bandwidth it's shared by students using it network out of the university and our collection visitors coming into the university. And the whole thing, the DS infrastructure, everything's inside the university. It's not bad in its present state but we need to look at a global audience, how do you scale it out, how do you balance it? And that's what we're looking at and it would've been almost impossible to meet the goals that we have, and the aspirations, and not to mention the cost. >> Okay so you're going to be at the summit, the Excel Center tomorrow right? What are you looking forward to there for us from a customer standpoint? >> I'm looking at service management, because a lot of our work, we've got a fantastic service desk and a fantastic team. So a lot of that is looking at service management, how to deliver effectively. As you rightly say Amazon is huge on innovation and things keep changing constantly so we need to keep track of how we deliver services, how do we make ourselves more nimble and more agile to deliver the services and add value. If you look at the OS stack, that's my favorite example, so you look at the OS stack you've got seven layers going up from physical then all the way to the application. You can almost read an organization in a similar way, so you got a physical level where you've got cabling and all the way to the people and presentation layer. So right now what we are doing is we are making sure we are focusing on the top level, focusing on the strategies, creating strategies, delivering that, rather than looking out for things that break. Looking out for things that operationally perhaps add value in another place. So that's where we would like to go. >> Anjenesh, thanks so much for coming on The Cube. >> Thank you >> It was a pleasure to have you. All right and thank you for watching, keep right there we'll be back with our next guest right after this short break. You're watching The Cube, from London at Amazon HQ, I call it HQ, we're here. Right back. (upbeat music)