Transformation and data capital
This article originally appeared in the IABM Journal, issue 105 which is available to view online here
We spoke to Dell EMC’s Thomas Burns (CTO Media & Entertainment) at the recent NAB Show about current opportunities and challenges in the industry and what the future holds.
What stood out as new or different at NAB Show – the trends and technologies that will have a major impact in the future of M&E?
The trend towards virtualization of server-based applications, so they can run on a converged compute stack anywhere – cloud or on-prem – is well established. Another is the migration from serial digital to IP networks. Whether in an on-prem data center, a colocation data center, or a private media cloud for high value content creation, the interesting new trend we are seeing is the adoption of enterprise IT workflows and infrastructure by media and entertainment facilities.
There have been cultural issues between broadcast engineers and IT guys, and like a lot of cultural change, sometimes it’s easy and sometimes it’s a real fight.
What we are finding is that there are advantages of scale, advantages of cost, and advantages of flexibility to adopting the infrastructure and workflows of enterprise IT. As an example, implementing the kind of change management techniques that enterprise IT has used for years is necessary for broadcasters as they move to a scale-out ‘entertainment services delivery’ model.
[bctt tweet=”What we are finding is that there are advantages of scale, advantages of cost, and advantages of flexibility to adopting the infrastructure and workflows of enterprise IT – IABM article on Transformation and data capital”]
Where does Dell EMC sit in Dell Technologies, and where are you focusing the business right now?
Dell EMC is one of seven strategically aligned businesses within Dell Technologies, alongside Dell, Pivotal, RSA, SecureWorks, Virtustream and VMware. The first overarching theme across all the businesses is transformation; that covers every aspect of operations across the four pillars: workforce, data center, security and digital transformation. The second is data capital, which we’ll get to later.
At Dell EMC, we are focused on core IT infrastructure – storage, servers and networking, and in terms of digital transformation, the M&E industry is already there. By virtualizing a broadcast playout application for instance, we are bringing the rest of the overarching Dell Technologies themes into play – data center transformation, workforce transformation, and security transformation.
Workforce transformation has major implications within M&E. A typical situation for a visual effects company when they are awarded the contract for the next 300 shots on a summer blockbuster picture is that they need to add 100 artists as quickly as possible. It’s a facility issue, a logistics issue, and a financial issue because they need to purchase and configure 100 workstations and find a place to put them. It’s an HR issue, because you need to get the artists aligned with their work visas, help them relocate, etc. In short, the data center transformation may largely be done in M&E, but the workforce transformation is ongoing.
We know how to virtualize broadcast playout and other server-based apps, and already do this via OEM agreements with most of the broadcast ISVs in this space. In fact, one of the happiest moments for me personally was at NAB Show this year, when a company that’s not currently a customer, with a large portfolio of applications, came by our booth at the end of the show and said, “Hey! We heard you are the guys we should do a deal with if we don’t want to be in the hardware business anymore.”
I just loved that because it meant that we were getting our messaging across at NAB Show, and the message was: “Why should you spend all of your resources developing specific hardware solutions when you can just virtualize it?” What I mean is – leave the hardware to us. That means broadcast application developers can focus on what they’re good at – applications – and not have to worry about upgrading somebody’s storage in five years, when the company that you originally spec’d for the storage is out of business.
The new focus of virtualization is the artists’ workstations. The cutting edge right now is people figuring out how to virtualize an Avid seat, an Adobe seat, etc. Some applications that are tightly tied to the hardware are more difficult to virtualize, and many of those companies are starting to offer subscription- based versions of their software, so you don’t have to capitalize, upgrade and maintain turnkey workstations.
Within Dell Technologies, we have the VMware expertise to fully virtualize graphics workstations – it’s an interesting challenge because it’s not just about the technology but other aspects too – for example, the licensing architecture and the plug-in ecosystem. But despite the challenges, workforce transformation by virtualizing or remoting the artists’ workstation is important in M&E because the industry is moving to globally collaborative production. Workforce transformation is one of the most important things happening in M&E, simply as a response to the dispersed nature of the labor force, and this is particularly so in VFX, animation, gaming and VR/AR.
The final pillar is security transformation, and I think we all know how important security is in this globally networked world. Broadcasters used to rely on physical security; only engineering had access to the central equipment room where the VTRs were, so we didn’t have to worry about unauthorized duplication as much. Nowadays we need a layered, prioritized approach to security. Implementing tightly firewalled media networks won’t help if a bad actor penetrates the control networks and maliciously turns off the AC, because that’ll shut down a broadcast plant faster than anything.
[bctt tweet=”Some applications that are tightly tied to the hardware are more difficult to virtualize, and many of those companies are starting to offer subscription-based versions of their software – IABM article on Transformation and data capital”]
Dell EMC talks about exploiting data capital to underpin the delivery of compelling content to audiences anytime, anywhere and on any device. Can you explain what data capital is and why it is so important?
Data capital is our major theme for 2018 into 2019. How that works in M&E, however, isn’t so different from what we have been telling people before this. Your data capital is the monetization value of what lies in your asset library. In broadcast especially, we have been working to implement that for several years, now that object storage is becoming cost-competitive with a tape library.
[bctt tweet=”Your data capital is the monetization value of what lies in your asset library…we have been working to implement that for several years, now that object storage is becoming cost-competitive with a tape library – IABM article on Transformation and data capital”]
The idea is that data has mass, just like gravity; the larger your data, the more attraction it has. And you need to get your data out of silos – data locked up in silos is very difficult to do anything with. The first thing we have to do is understand that putting all of your data in a data lake, and bringing the applications to the data, is the way to monetize it. Otherwise, moving these gigantic assets that have so much mass would take weeks to migrate out of one silo and into another, e.g. for analytics. Put media in a central data lake, get it out of the silos, and you can bring the applications to the data. Video data has so much mass that it actually attracts applications, and facilities are realizing that it’s financially ruinous to do it the other way around.
I don’t like to see people idle just because they are copying a file from one silo to another. When you have assets in a data lake, you can use search tools, add metadata, and unlock the value of the capital assets that you own. That may be a new message for enterprise IT, but adding metadata to your existing library so you can better monetize it is something that a number of rights holders and service providers in the M&E space have been trying to do for a while. There have been a lot of early indicators, but no one has the gold ring, yet.
Everybody is working on it because they are realizing that is the way they want to go.
Artificial intelligence – which was very buzzy at NAB Show this year – shows incredible potential, but at present the only thing that I can see AI/ML/DL being used for is dividing the entire internet into ‘this is a picture of a cat, this is not a picture of a cat.’ That may be where we are at today, but things are moving fast; I have seen demonstrations that have automated facial recognition, recognition of team numbers on players’ jerseys, creating synthetic camera angles within a video volume; it is poised to go incredibly fluid, and there will be outcomes that we can’t yet fathom.
[bctt tweet=”Artificial intelligence – which was very buzzy at NAB Show this year – shows incredible potential…it is poised to go incredibly fluid, and there will be outcomes that we can’t yet fathom – IABM article on Transformation and data capital”]
Have you ever heard the quote that whenever we try and predict the future, we overestimate the short-term impact, and vastly underestimate what is going to change in the long run? I think we are at that stage with AI. But using AI to scan and tag the contents of your library holds enormous promise: to unlock capital assets from your film and tape library and add metadata to them. Then we can start to realize the value, or potential value, of the data capital.
Of course, there are companies that are promising much, much better than ‘this is a cat, this is not a cat.’ We are talking to companies who are taking this technology that exists at the algorithm level, and turning it into a product that we can bundle with a bunch of GPUs in a Dell server and give people a turnkey appliance which can scan through their library and produce actionable valuations on their media assets.
We are close but not quite there yet.
According to your NAB Show presentation, on-premise storage is still the most common in M&E. Will the cloud ever take over and if so, what needs to happen, or will local storage always have a place to play?
We are going to need on-premise storage for content creation because the applications aren’t yet object- native. Object storage isn’t really a performance technology to begin with, plus you need a file-to-object gateway of some kind, which increases latency. I think that you’re going to have on-prem storage for all the high-throughput, low latency workloads such as editing, visual effects, gaming – things where you need high- throughput single-stream performance. You know, if you’re in the color grading suite working on the latest blockbuster feature, full aperture DPX 4K at 24fps, you’re looking at over 1 GB/s sustained and that is difficult to achieve from the cloud.
In reality you need both on-prem and cloud storage – this was an important part of our NAB announcements this year. We’ve completed a reorganization of Isilon and ECS, putting both our file and object teams together as one product line. We call the new team ‘Unstructured Data Solutions’ – FYI anything that is not ‘block’ storage is ‘unstructured’, which by the way is the fastest- growing segment of the storage market. The product teams, engineering teams and the sales teams are all together, so that way we are working with a unified file and object solution. That differentiates us from every other storage vendor on the floor at NAB Show, in that we develop, sell, support and work with files and objects as two ends of the spectrum. Until all of the applications are object native, and until the various network protocols are more performance-oriented, I am certain that we will need at one end of the spectrum, on-prem file, to work seamlessly with the other end of the spectrum which is a private media cloud for, or a hybrid cloud which many of our larger customers are implementing.
So putting it all in the cloud is not always the best solution then?
We are seeing that, once people have figured out their workflow, once they have six months of historical data, and they can accurately predict the spikes and troughs, then it’s about bringing that workflow back on-prem for cost control; we are seeing large household names bringing their data back on-prem. The public cloud is great when you need to rapidly gin up a workflow, and try out a business model, and play around with things, and you don’t know where the data is going to spike, so you don’t know what infrastructure you need to buy – and even if you did know when to buy, it’s going to be 3-4 weeks before you can get it on-prem. The ability to just start something right now is fantastic, and that’s why I love the public cloud. But once you have a known workflow, and you have six months’ worth of data analysis on your traffic pattern, then you can crunch the numbers and figure out that it’s much less than the cost of public cloud to bring that same workflow on-prem – and that’s total cost of ownership.
[bctt tweet=”It’s about bringing that workflow back on-prem for cost control; we are seeing large household names bringing their data back on-prem. – IABM article on Transformation and data capital”]
Without giving away any trade secrets, what else is on Dell EMC’s to do list for the next 2-5 years – what should we be preparing for that isn’t much talked about yet?
First of all, given the fact that we are now part of Dell we can take advantage of Dell’s market-leading servers.
Dell is also very strong in networking – it has a huge practice in software-defined networking and spine and leaf architectures from its telco practice, and we will be looking to leverage this. Spine and leaf is one of the most exciting things that is coming down the pipeline partly because the market is ripe for us to take advantage of it, and partly because there is all this expertise on the Dell side that we can leverage.
And then, we’re looking at integrating search, integrating file to object, and object to file, making sure that there are no bottlenecks in that continuum all the way from on-prem high-throughput applications, to archive and disaster recovery and collaboration that we get from the private cloud or the object store.
