SVG Sit-Down: Spectra Logic CTO Matt Starr on the Rise of the Cloud and Why It’s a Great Time To Be in the Tape Business

The cloud and tape promise storage options for content owners of every size

Sports-video–content producers are facing bigger questions than ever in storing, archiving, and managing valuable assets. As the explosion of cloud-based storage and virtualized workflows changes the M&E market, such organizations must make tough decisions in laying out technology roadmaps. In many cases, sports networks and leagues are developing hybrid private/public-cloud infrastructures. Nonetheless, tape libraries still represent the cheapest option in most cases.

A 27-year vet of Spectra Logic, Matt Starr took over the role of CTO in 2000 and now oversees the company’s efforts in high-performance computing, private-cloud, and other vertical markets.

Matt Starr: “If you’re storing massive amounts of data and you’re not on tape, you’re going to go out of business.”

SVG sat down with Starr to discuss the current state of tape-based storage in the sports-video market, the future of cloud storage and virtualized workflows, why he believes disk storage capacities are not growing quickly enough to keep up with the expansion of data, the importance of genetic diversity in archiving assets, and how Spectra Logic is evolving its portfolio with products like the BlackPearl Converged Storage System and Arctic Blue disk-based storage to serve the rapidly changing M&E market.

How do you see the explosion of cloud storage and virtualized workflows impacting the tape-based–storage industry?
The important thing to note is, the cloud is not made up of pixie dust and cotton candy; it’s made up of IT equipment. You can store your data locally, or you can store your data in the cloud, but your data, someday, is going to end up on standardized IT equipment: a disk drive or a tape library. We see the same thing happening across all our verticals: media and entertainment, high-performance computing (HPC), federal government, video surveillance, and general IT. When you have data sets that are large, cloud storage tends to get too expensive. Cloud playout, on the other hand, which is what a lot of these providers like Telstra in Australia are trying to do, works very well because you don’t have to [implement] all the playout services but you still own your content and you just decide what should be played to air separately [in the cloud]. We certainly see a trend moving towards more of a hybrid cloud than a full-on cloud [model]. And I agree with that model because, in most cases, especially if it’s not your area of expertise, it allows you to let somebody else do the work for you.

What are the greatest challenges for an M&E operation looking to embrace cloud-based storage?
I’ve seen customers who have gone to Amazon Web Services for storage and other services and then found out that, despite the pricing that was listed, the fine print included some massive upcharges. My advice as a consumer is to read the fine print and know how much it will cost to change or get off that cloud. If you need to pull all of your workflows and operations back in-house, what will that cost you?

There’s a little bit of a lack of focus in the cloud-services market. Right now, there are about 13 or 14 major cloud–storage-service providers. You have the major names like EMC, HP, Amazon, Microsoft, and Google. Then you go outside of [that group], and there are many others. I don’t want to say there are too many [cloud-services vendors], but there is a feeling that everyone’s getting into that space. It’s going to thin out at some point, which will help clarify the market.

How important is it to consider genetic diversity in developing a storage scheme for your valuable assets?
It depends on if the asset is irreplaceable or not. There are assets, through a transcode or something else, that can be recomputed. So do I need to store that copy twice [in different formats]? Probably not. Then there are assets that can’t be replaced, and that’s where you really want to apply that genetic diversity.

One thing to consider about genetic diversity in the public cloud is, you don’t always know what [the cloud vendor is] using and usually they won’t tell you. So you could be making a copy on LTO tape locally and thinking that you’re diversifying by putting it in the cloud. Well, I know enough cloud vendors, and I know some that use LTO tape. So is that genetic diversity? No. If I were doing that and I wanted it for genetic diversity, I’d make sure the cloud provider was a little more forthright on the [storage] technology being used. With Amazon, you can see what level of service you have, but that’s not always the case.

When does it make sense for an M&E organization to store content on tape vs. disk vs. the cloud vs. a hybrid model?
If you’re storing massive amounts of data and you’re not on tape, you’re going to go out of business. You can dislike this technology, but your competitor is going find a way to use it to their advantage. Cost-wise, Flash is usually between 10 times and 30 times what tape is. Disk is usually between four times and 10 times what tape is. Tape is down to 1¢ or 2¢ per gig on the shelf and, managed, at probably 5¢ or 6¢ per gig. My general rule is, if you have half a petabyte of data, that’s still a viable candidate for a disk and a backup. When you cross that or you have 500 TB and you’re growing rapidly, that’s when you want to start thinking about [a tape-based] archive.

Also, [if you’re] a customer spinning a few hundred terabytes [on disk], make sure your data is backed up. Where do you want to back it up then? If you want to go cloud, 500 TB will cost you several hundred thousand dollars in five years. You could buy a lot of [equipment] for that few hundred thousand.

Can you detail Spectra Logic’s BlackPearl Converged Storage System? What has the market response been? Why do you believe it is integral to the company’s success?
BlackPearl solves the problem of costly and complex approaches to digital preservation by combining NAS- and S3-based interfaces with multiple storage targets into a simple and affordable solution. Designed for numerous concurrent workflows, BlackPearl reduces the need for expensive third-party data movers by integrating Spectra S3 with a range of certified clients and simple file movers.

Since its introduction in 2013, BlackPearl has had tremendous momentum, as organizations seek to simply and affordably preserve ever-increasing amounts of data forever. Current BlackPearl customers span genomics and healthcare, general IT, high-performance computing, media and entertainment, research and education, and video surveillance. Customers include British Film Institute; WCBS-TV; Avid Technologies; Tiger Technology; CatDV; IPV; and Empress Media Asset Management. twofour54, one of the largest media and entertainment-services companies in the United Arab Emirates, [is using] Spectra BlackPearl coupled with the T120 tape library.

BlackPearl is at the heart of Spectra’s hybrid storage ecosystem strategy and the gateway to tape, disk, and public and hybrid cloud. Scalable to exabyte capacities at pennies per gigabyte, BlackPearl is especially relevant to our media and entertainment customers facing increased camera resolutions, higher frame rates, and continued growth in the output of movies, news, sports, and regular programming. Moving from an HD format at 24 frames per second to 4K format at 60 fps requires roughly 30 times the storage, and most content is now being stored forever.

The need for affordable, scalable storage will only grow: the amount of data stored in the “digital universe” will reach 17 zettabytes, or 1 sextillion [10²¹] bytes by 2025. BlackPearl is fundamental to smart storage solutions in today’s environment and will serve as the basis for our next-wave products, which will debut later this year.

As production and distribution costs drop, we’re seeing massive growth in creation of niche sports content, such as amateur and collegiate Olympic sports? How can these properties justify long-term storage, archive, and management of their content. Or can they?
In the sports market, [with] the Motocross channel or Speed network or another network like that, playout and edit doesn’t happen in their own facilities. Instead, they’re going to a third-party company and having their edit and some of their playout done by somebody there. Is that a cloud provider? I would say yes. They are amortizing an extra 10 or 15 channels over their own infrastructure, and then they can pay for their infrastructure. You’re going to end up with that kind of model more often than you think. You’ll see a model where the smallest [organizations] will be on a shared infrastructure that is leased and the largest [organizations] will own their infrastructure.

How do you see the M&E market evolving in the next few years, and how is Spectra Logic looking to serve those changing needs?
The M&E market is no longer the fat business it was, where they had a lot of money and they could buy the best of everything. That is long gone, and the reason is the number of channels and content being streamed today and that the advertising dollars haven’t kept up. You have thousands of channels now vs. 50 channels just a few years ago.

The organizations who are smarter in reducing their infrastructure will be very [successful]. I think that’s where cloud-based playout and things like that may bring value. [It’s] where you do your core-value pieces, maybe editing and things like that, and you offload other pieces to something that can amortize a whole bunch of equipment across 100 people. In the next 12 months, I think you will continue to see people figure out how to do things smarter and better. And the people who write the software that does the job more efficiently are going to be the winners. You’ll see companies like Front Porch Digital, who I understand has shrunk [significantly] … since [its] sale to Oracle a little over a year ago, fade away.

Beyond the next 12 months is where it gets interesting. I question whether the [viewership] is enough to [justify] the content. For example, I’m watching a BBC show on Netflix called Plane Resurrection. It’s a BBC show about people who restore World War II aircraft. There are probably thousands of people that watch the show, but it certainly isn’t tens of thousands. What is the cost to make that vs. the return on investment for an eight-part series? In the next five years, you may see people stepping back and saying where is the value in this content? Is it worth trying to remonetize some old sporting event that isn’t providing any value?

How will Spectra Logic adapt to this rapidly changing M&E market?
First, things must go faster. For tape and archive in general, we have to [reduce] time-to-video. That’s one thing we’re focusing on. The other thing we’re certainly focusing on is the hybrid cloud. We think that content distribution — getting content anywhere, content sharing — is going to be extremely important. We’ve looked at a Netflix-like model for our product and believe, if we apply a Netflix model to our product and then do rights management, we could have a very shareable content system for people to deploy. We’re talking about a company that wants to do shared aspects across the U.S. or across the world. We want to be able to provide both true-cloud storage and local-cloud storage and have that data accessible from anywhere in the world.

Some have said for decades that tape is dead, but you have to consider that tape is still the cheapest [storage option] on the market today. I’m very happy to say that we’re actually accelerating a tape-storage roadmap. If you look at LTO.org, our website, or others, you’ll see that delivery date and capacities are starting to kick up to where they should be. It’s really nice, in my opinion, to be past the time when disk was creeping in on tape’s ecosystem. Tape is now accelerating, and it’s a good place and a good time to be with large tape systems.