By Randi Altman

Every year in our special Storage Edition, we poll those who use storage and those who make storage. This year is no different. The users we’ve assembled for our latest offering weigh in on how they purchase gear, how they employ storage and cloud-based solutions. Storage makers talk about what’s to come from them, how AI and ML are affecting their tools, NVMe growth and more.

Enjoy…

Periscope Post & Audio, GM, Ben Benedetti

Periscope Post & Audio is a full-service post company with facilities in Hollywood and Chicago’s Cinespace. Both facilities provide a range of sound and picture finishing services for TV, film, spots, video games and other media.

What types of storage are you using for your workflows?
For our video department, we have a large, high-speed Quantum media array supporting three color bays, two online edit suites, a dailies operation, two VFX suites and a data I/O department. The 15 systems in the video department are connected via 16GB fiber.

For our sound department, we are using an Avid Nexis System via 6e Ethernet supporting three Atmos mix stages, two sound design suites, an ADR room and numerous sound-edit bays. All the CPUs in the facility are securely located in two isolated machine rooms (one for video on our second floor and one for audio on the first). All CPUs in the facility are tied via an IHSE KVM system, giving us incredible flexibility to move and deliver assets however our creatives and clients need them. We aren’t interested in being the biggest. We just want to provide the best and most reliable services possible.

Cloud versus on-prem – what are the pros and cons?
We are blessed with a robust pipe into our facility in Hollywood and are actively discussing with our engineering staff about using potential cloud-based storage solutions in the future. We are already using some cloud-based solutions for our building’s security system and CCTV systems as well as the management of our firewall. But the concept of placing client intellectual property in the cloud sparks some interesting conversations.We always need immediate access to the raw footage and sound recordings of our client productions, so I sincerely doubt we will ever completely rely on a cloud-based solution for the storage of our clients’ original footage. We have many redundancy systems in place to avoid slowdowns in production workflows. This is so critical. Any potential interruption in connectivity that is beyond our control gives me great pause.

How often are you adding or upgrading your storage?
Obviously, we need to be as proactive as we can so that we are never caught unready to take on projects of any size. It involves continually ensuring that our archive system is optimized correctly and requires our data management team to constantly analyze available space and resources.

How do you feel about the use of ML/AI for managing assets?
Any AI or ML automated process that helps us monitor our facility is vital. Technology advancements over the past decade have allowed us to achieve amazing efficiencies. As a result, we can give the creative executives and storytellers we service the time they need to realize their visions.

What role might the different tiers of cloud storage play in the lifecycle of an asset?
As we have facilities in both Chicago and Hollywood, our ability to take advantage of Google cloud-based services for administration has been a real godsend. It’s not glamorous, but it’s extremely important to keeping our facilities running at peak performance.

The level of coordination we have achieved in that regard has been tremendous. Those low-tiered storage systems provide simple and direct solutions to our administrative and accounting needs, but when it comes to the high-performance requirements of our facility’s color bays and audio rooms, we still rely on the high-speed on-premises storage solutions.

For simple archiving purposes, a cloud-based solution might work very well, but for active work currently in production … we are just not ready to make that leap … yet. Of course, given Moore’s Law and the exponential advancement of technology, our position could change rapidly. The important thing is to remain open and willing to embrace change as long as it makes practical sense and never puts your client’s property at risk.

Panasas, Storage Systems Engineer, RW Hawkins

Panasas offers a scalable high-performance storage solution. Its PanFS parallel file system, delivered on the ActiveStor appliance, accelerates data access for VFX feature production, Linux-based image processing, VR/AR and game development, and multi-petabyte sized active media archives.

What kind of storage are you offering, and will that be changing in the coming year?
We just announced that we are now shipping the next generation of the PanFS parallel file system on the ActiveStor Ultra turnkey appliance, which is already in early deployment with five customers.

This new system offers unlimited performance scaling in 4GB/s building blocks. It uses multi-tier intelligent data placement to maximize storage performance by placing metadata on low-latency NVMe SSDs, small files on high IOPS SSDs and large files on high-bandwidth HDDs. The system’s balanced-node architecture optimizes networking, CPU, memory and storage capacity to prevent hot spots and bottlenecks, ensuring high performance regardless of workload. This new architecture will allow us to adapt PanFS to the ever-changing variety of workloads our customers will face over the next several years.

Are certain storage tiers more suitable for different asset types, workflows, etc.?
Absolutely. However, too many tiers can lead to frustration around complexity, loss of productivity and poor reliability. We take a hybrid approach, whereby each server has multiple types of storage media internal to one server. Using intelligent data placement, we put data on the most appropriate tier automatically. Using this approach, we can often replace a performance tier and a tier two active archive with one cost-effective appliance. Our standard file-based client makes it easy to gateway to an archive tier such as tape or an object store like S3.

What do you see are the big technology trends that can help storage for M&E? ML? AI?
AI/ML is so widespread, it seems to be all encompassing. Media tools will benefit greatly because many of the mundane production tasks will be optimized, allowing for more creative freedom. From a storage perspective, machine learning is really pushing performance in new directions; low latency and metadata performance are becoming more important. Large amounts of unstructured data with rich metadata are the norm, and today’s file systems need to adapt to meet these requirements.

How has NVMe advanced over the past year?
Everyone is taking notice of NVMe; it is easier than ever to build a fast array and connect it to a server. However, there is much more to making a performant storage appliance than just throwing hardware at the problem. My customers are telling me they are excited about this new technology but frustrated by the lack of scalability, the immaturity of the software and the general lack of stability. The proven way to scale is to build a file system on top of these fast boxes and connect them into one large namespace. We will continue to augment our architecture with these new technologies, all the while keeping an eye on maintaining our stability and ease of management.

Do you see NAS overtaking SAN for larger work groups? How about cloud taking on some of what NAS used to do?
Today’s modern NAS can take on all the tasks that historically could only be done with SAN. The main thing holding back traditional NAS has been the client access protocol. With network-attached parallel clients, like Panasas’ DirectFlow, customers get advanced client caching, full POSIX semantics and massive parallelism over standard ethernet.

Regarding cloud, my customers tell me they want all the benefits of cloud (data center consolidation, inexpensive power and cooling, ease of scaling) without the vendor lock-in and metered data access of the “big three” cloud providers. A scalable parallel file system forms the core of a private cloud model that yields the benefits without the drawbacks. File-based access to the namespace will continue to be required for most non-web-based applications.

Goldcrest Post, New York, Technical Director, Ahmed Barbary

Goldcrest Post is an independent post facility, providing solutions for features, episodic TV, docs, and other projects. The company provides editorial offices, on-set dailies, picture finishing, sound editorial, ADR and mixing, and related services.

What types of storage are you using for your workflows?
Storage performance in the post stage is tremendously demanding. We are using multiple SAN systems in office locations that provide centralized storage and easy access to disk arrays, servers, and other dedicated playout applications to meet storage needs throughout all stages of the workflow.

While backup refers to duplicating the content for peace of mind, short-term retention, and recovery, archival signifies transferring the content from the primary storage location to long-term storage to be preserved for weeks, months, and even years to come. Archival storage needs to offer scalability, flexible and sustainable pricing, as well as accessibility for individual users and asset management solutions for future projects.

LTO has been a popular choice for archival storage for decades because of its affordable, high-capacity solutions with low write/high read workloads that are optimal for cold storage workflows. The increased need for instant access to archived content today, coupled with the slow roll-out of LTO-8, has made tape a less favorable option.

Cloud versus on-prem – what are the pros and cons?
The fact is each option has its positives and negatives, and understanding that and determining how both cloud and on-premises software fit into your organization are vital. So, it’s best to be prepared and create a point-by-point comparison of both choices.

When looking at the pros and cons of cloud vs. on-premises solutions, everything starts with an understanding of how these two models differ. With a cloud deployment, the vendor hosts your information and offers access through a web portal. This enables more mobility and flexibility of use for cloud-based software options. When looking at an on-prem solution, you are committing to local ownership of your data, hardware, and software. Everything is run on machines in your facility with no third-party access.

How often are you adding or upgrading your storage?
We keep track of new technologies and continuously upgrade our systems, but when it comes to storage, it’s a huge expense. When deploying a new system, we do our best to future-proof and ensure that it can be expanded.

How do you feel about the use of ML/AI for managing assets?
For most M&E enterprises, the biggest potential of AI lies in automatic content recognition, which can drive several path-breaking business benefits. For instance, most content owners have thousands of video assets.

Cataloging, managing, processing, and re-purposing this content typically requires extensive manual effort. Advancements in AI and ML algorithms have
now made it possible to drastically cut down the time taken to perform many of these tasks. But there is still a lot of work to be done — especially as ML algorithms need to be trained, using the right kind of data and solutions, to achieve accurate results.

What role might the different tiers of cloud storage play in the lifecycle of an asset?
Data sets have unique lifecycles. Early in the lifecycle, people access some data often, but the need for access drops drastically as the data ages. Some data stays idle in the cloud and is rarely accessed once stored. Some data expires days or months after creation, while other data sets are actively read and modified throughout their lifetimes.

Rohde & Schwarz, Product Manager, Storage Solutions, Dirk Thometzek

Rohde & Schwarz offers broadcast and media solutions to help companies grow in media production, management and delivery in the IP and wireless age.

What kind of storage are you offering, and will that be changing in the coming year?
The industry is constantly changing, so we monitor market developments and key demands closely. We will be adding new features to the R&S SpycerNode in the next few months that will enable our customers to get their creative work done without focusing on complex technologies. The R&S SpycerNode will be extended with JBODs, which will allow seamless integration with our erasure coding technology, guaranteeing complete resilience and performance.

Are certain storage tiers more suitable for different asset types, workflows, etc.?
Each workflow is different, so, consequently, there is almost no system alike. The real artistry is to tailor storage systems according to real requirements without over-provisioning hardware or over-stressing budgets. Using different tiers can be very helpful to build effective systems, but they might introduce additional difficulties to the workflows if the system isn’t properly designed.

Rohde & Schwarz has developed R&S SpycerNode in a way that its performance is linear and predictable. Different tiers are aggregated under a single namespace, and our tools allow seamless workflows while complexity remains transparent to the users.

What do you see are the big technology trends that can help storage for M&E? ML? AI?
Machine learning and artificial intelligence can be helpful to automate certain tasks, but they will not replace human intervention in the short term. It might not be helpful to enrich media with too much data because doing so could result in imprecise queries that return far too much content.

However, clearly defined changes in sequences or reoccurring objects — such as bugs and logos — can be used as a trigger to initiate certain automated workflows. Certainly, we will see many interesting advances in the future.

How has NVMe advanced over the past year?
NVMe has very interesting aspects. Data rates and reduced latencies are admittedly quite impressive and are garnering a lot of interest. Unfortunately, we do see a trend inside our industry to be blinded by pure performance figures and exaggerated promises without considering hardware quality, life expectancy or proper implementation. Additionally, if well-designed and proven solutions exist that are efficient enough, then it doesn’t make sense to embrace a technology just because it is available.

R&S is dedicated to bringing high-end devices to the M&E market. We think that reliability and performance build the foundation for user-friendly products. Next year, we will update the market on how NVMe can be used in the most efficient way within our products.

Do you see NAS overtaking SAN for larger work groups? How about cloud taking on some of what NAS used to do?
We definitely see a trend away from classic Fibre Channel to Ethernet infrastructures for various reasons. For many years, NAS systems have been replacing central storage systems based on SAN technology for a lot of workflows. Unfortunately, standard NAS technologies will not support all necessary workflows and applications in our industry. Public and private cloud storage systems play an important role in overall concepts, but they can’t fulfil all necessary media production requirements or ease up workflows by default. Plus, when it comes to subscription models, [sometimes there could be unexpected fees]. In fact, we do see quite a few customers returning to their previous services, including on-premises storage systems such as archives.

When it comes to the very high data rates necessary for high-end media productions, NAS will relatively quickly reach its technical limits. Only block-level access can deliver the reliable performance necessary for uncompressed productions at high frame rates.

That does not necessarily mean Fibre Channel is the only solution. The R&S SpycerNode, for example, features a unified 100Gb/s Ethernet backbone, wherein clients and the redundant storage nodes are attached to the same network. This allows the clients to access the storage over industry-leading NAS technology or native block level while enabling true flexibility using state-of-the-art technology.

MTI Film, CEO, Larry Chernoff

Hollywood’s MTI Film is a full-service post facility, providing dailies, editorial, visual effects, color correction, and assembly for film, television, and commercials.

What types of storage are you using for your workflows?
MTI uses a mix of spinning and SSD disks. Our volumes range from 700TB to 1000TB and are assigned to projects depending on the volume of expected camera files. The SSD volumes are substantially smaller and are used to play back ultra-large-resolution files, where several users are using the file.

Cloud versus on-prem — what are the pros and cons?
MTI only uses on-prem storage at the moment due to the real-time, full-resolution nature of our playback requirements. There is certainly a place for cloud-based storage but, as a finishing house, it does not apply to most of our workflows.

How often are you adding or upgrading your storage?
We are constantly adding storage to our facility. Each year, for the last five, we’ve added or replaced storage annually. We now have approximately 8+ PB, with plans for more in the future.

How do you feel about the use of ML/AI for managing assets?
Sounds like fun!

What role might the different tiers of cloud storage play in the lifecycle of an asset?
For a post house like MTI, we consider cloud storage to be used only for “deep storage” since our bandwidth needs are very high. The amount of Internet connectivity we would require to replicate the workflows we currently have using on-prem storage would be prohibitively expensive for a facility such as MTI. Speed and ease of access is critical to being able to fulfill our customers’ demanding schedules.

OWC,Founder/CEO, Larry O’Connor

OWC offers storage, connectivity, software, and expansion solutions designed to enhance, accelerate, and extend the capabilities of Mac- and PC-based technology. Their products range from the home desktop to the enterprise rack to the audio recording studio to the motion picture set and beyond.

What kind of storage are you offering, and will that be changing in the coming year?
OWC will be expanding our Jupiter line of NAS storage products in 2020 with an all new external flash base array. We will also be launching the OWC ThunderBay Flex 8, a three-in-one Thunderbolt 3 storage, docking, and PCIe expansion solution for digital imaging, VFX, video production, and video editing.

Are certain storage tiers more suitable for different asset types, workflows etc?
Yes. SSD and NVMe are better for on-set storage and editing. Once you are finished and looking to archive, HDD are a better solution for long term storage.

What do you see are the big technology trends that can help storage for M&E? ML? AI?
We see U.2 SSDs as a trend that can help storage in this space. Also, solutions that allow the use of external docking of U.2 across different workflow needs.

How has NvME advanced over the past year?
We have seen NVMe technology become higher in capacity, higher in performance, and substantially lower in power draw. Yet even with all the improving performance, costs are lower today versus 12 months ago. SSD and NVMe are better for on-set storage and editing. Once you are finished and looking to archive, HDD are a better solution for long term storage.

Do you see NAS overtaking SAN for larger work groups? How about cloud taking on some of what NAS used to do?
I see both still having their place — I can’t speak to if one will take over the other. SANs provide other services that typically go hand in hand with M&E needs.

As for cloud, I can see some more cloud coming in, but for M&E on-site needs, it just doesn’t compete anywhere near with what the data rate demand is for editing, etc. Everything independently has its place.

EditShare, VP of Product Management, Sunil Mudholkar

EditShare offers a range of media management solutions, from ingest to archive with a focus on media and entertainment.

What kind of storage are you offering and will that be changing in the coming year?
EditShare currently offers RAID and SSD, along with our nearline Sata HDD-based storage. We are on track to deliver NVMe- and cloud-based solutions in the first half of 2020. The latest major upgrade of our file system and management console, EFS2020, enables us to migrate to emerging technologies, including cloud deployment and using NVMe hardware.

EFS can manage and use multiple storage pools, enabling clients to use the most cost-effective tiered storage for their production, all while keeping that single namespace.

Are certain storage tiers more suitable for different asset types, workflows etc?
Absolutely. It’s clearly financially advantageous to have varying performance tiers of storage that are in line with the workflows the business requires. This also extends to the cloud, where we are seeing public cloud-based solutions augment or replace both high-performance and long-term storage needs. Tiered storage enables clients to be at their most cost effective by including parking storage and cloud storage for DR, while keeping SSD and NVME storage ready and primed for their high-end production.

What do you see are the big technology trends that can help storage for M&E? ML? AI?
AI and ML have somewhat of an advantage for storage when it comes to things like algorithms that are designed to automatically move content between storage tiers to optimize costs. This has been commonplace in the distribution side of the ecosystem for a long time with CDNs. ML and AI have a great ability to impact the Opex side of asset management and metadata by helping to automate very manual, repetitive data entry tasks through audio and image recognition, as an example.

AI can also assist by removing mundane human-centric repetitive tasks, such as logging incoming content. AI can assist with the growing issue of unstructured and unmanaged storage pools, enabling the automatic scanning and indexing of every piece of content located on a storage pool.

How has NVMe advanced over the past year?
Like any other storage medium, when it’s first introduced there are limited use cases that make sense financially, and only a certain few can afford to deploy it. As the technology scales and changes in form factor, and pricing becomes more competitive and inline with other storage options, it then can become more mainstream. This is what we are starting to see with NVMe.

Do you see NAS overtaking SAN for larger work groups? How about cloud taking on some of what NAS used to do?
Yes, NAS has overtaken SAN. It’s easier technology to deal with — this is fairly well acknowledged. It’s also easier to find people/talent with experience in NAS. Cloud will start to replace more NAS workflows in 2020, as we are already seeing today. For example, our ACL media spaces project options within our management console were designed for SAN clients migrating to NAS. They liked the granular detail that SAN offered, but wanted to migrate to NAS. EditShare’s ACL enables them to work like a SAN but in a NAS environment.

Zoic Studios CTO Saker Klippsten

Zoic Studios is an Emmy-winning VFX company based in Culver City, California, with sister offices in Vancouver and NYC. It creates computer-generated special effects for commercials, films, television and video games.

What types of projects are you working on?
We work on a range of projects for series, film, commercial and interactive games (VR/AR). Most of the live-action projects are mixed with CG/VFX and some full-CG animated shots. In addition, there is typically some form of particle or fluid effects simulation going on, such as clouds, water, fire, destruction or other surreal effects.

What types of storage are you using for those workflows?
Cryogen – Off-the-shelf tape/disk/chip. Access time > 1 day. Mostly tape-based and completely offline, which requires human intervention to load tapes or restore from drives.
Freezing – Tape robot library. Access time < .5 day. Tape-based and in the robot. This does not require intervention.Cold – Spinning disk. Access time — slow (online). Disaster recovery and long-term archiving.
Warm – Spinning disk. Access time — medium (online). Data that needs to still be accessed promptly and transferred quickly (asset depot).
Hot – Chip-based. Access time — fast (online). SSD generic active production storage.
Blazing – Chip-based. Access time — uber fast (online). NVMe dedicated storage for 4K and 8K playback, databases and specific simulation workflows.

Cloud versus on-prem – what are the pros and cons?
The great debate! I tend to not look at it as pro vs. con, but where you are as a company. Many factors are involved and there is no one size that fits all, as many are led to believe, and neither cloud or on-prem alone can solve all your workflow and business challenges.

There are workflows that are greatly suited for the cloud and others that are potentially cost prohibitive for a number of reasons, such as the size of the data set being generated. Dynamics Cache Simulations are a good example, which can quickly generate tens of TBs or sometimes hundreds of TBs. If the workflow requires you to transfer this data on premises for review, it could take a very long time. Other workflows such as 3D CG-generated data can take better advantage of the cloud. They typically have small source file payloads that need to be uploaded and then only require final frames to be downloaded, which is much more manageable. Depending on the size of your company and level of technical people on hand, the cloud can be a problem

What triggers buying more storage in your shop?
Storage tends to be one of the largest and most significant purchases at many companies. End users do not have a clear concept of what happens at the other end of the wire from their workstation.

All they know is that there is never enough storage and it’s never fast enough. Not investing in the right storage can not only be detrimental to the delivery and production of a show, but also to the mental focus and health of the end users. If artists are constantly having to stop and clean up/delete, it takes them out of their creative rhythm and slows down task completion.

If the storage is not performing properly and is slow, this will not only have an impact on delivery, but the end user might be afraid they are being perceived as being slow. So what goes into buying more storage? What type of impact will buying more storage have on the various workflows and pipelines? Remember, if you are a mature company you are buying 2TB of storage for every 1TB required for DR purposes, so you have a complete up-to-the-hour backup.

Do you see ML/AI as important to your content strategy?
We have been using various layers of ML and heuristics sprinkled throughout our content workflows and pipelines. As an example, we look at the storage platforms we use to understand what’s on our storage, how and when it’s being used, what it’s being used for and how it’s being accessed. We look at the content to see what it contains and its characteristics. What are the overall costs to create that content? What insights can we learn from it for similarly created content? How can we reuse assets to be more efficient?

Dell Technologies, CTO, Media & Entertainment, Thomas Burns

Dell offers technologies across workstations, displays, servers, storage, networking and VMware, and partnerships with key media software vendors to provide media professionals the tools to deliver powerful stories, faster.

What kind of storage are you offering, and will that be changing in the coming year?
Dell Technologies offers a complete range of storage solutions from Isilon all-flash and disk-based scale-out NAS to our object storage, ECS, which is available as an appliance or a software-defined solution on commodity hardware. We have also developed and open-sourced Pravega, a new storage type for streaming data (e.g. IoT and other edge workloads), and continue to innovate in file, object and streaming solutions with software-defined and flexible consumption models.

Are certain storage tiers more suitable for different asset types, workflows etc?
Intelligent tiering is crucial to building a post and VFX pipeline. Today’s global pipelines must include software that distinguishes between hot data on the fastest tier and cold or versioned data on less performant tiers, especially in globally distributed workflows. Bringing applications to the media rather than unnecessarily moving media into a processing silo is the key to an efficient production.

What do you see are the big technology trends that can help storage for M&E? ML? AI?
New developments in storage class memory (SCM) — including the use of carbon nanotubes to create a nonvolatile, standalone memory product with speeds rivaling DRAM without needing battery backup — have the potential to speed up media workflows and eliminate AI/ML bottlenecks. New protocols such as NVMe allow much deeper I/O queues, overcoming today’s bus bandwidth limits.

GPUDirect enables direct paths between GPUs and network storage, bypassing the CPU for lower latency access to GPU compute — desirable for both M&E and AI/ML applications. Ethernet mesh, a.k.a. Leaf/Spine topologies, allow storage networks to scale more flexibly than ever before.

How has NVMe advanced over the past year?
Advances in I/O virtualization make NVMe useful in hyper-converged infrastructure, by allowing different virtual machines (VMs) to share a single PCIe hardware interface. Taking advantage of multi-stream writes, along with vGPUs and vNICs, allows talent to operate more flexibly as creative workstations start to become virtualized.

Do you see NAS overtaking SAN for larger work groups? How about cloud taking on some of what NAS used to do?
IP networks scale much better than any other protocol, so NAS allows on-premises workloads to be managed more efficiently than SAN. Object stores (the basic storage type for cloud services) support elastic workloads extremely well and will continue to be an integral part of public, hybrid and private cloud media workflows.

ATTO, Manager, Products Group, Peter Donnelly

ATTO network and storage connectivity products are purpose-made to support all phases of media production, from ingest to final archiving. ATTO offers an ecosystem of high-performance connectivity adapters, network interface cards and proprietary software.

What kind of storage are you offering, and will that be changing in the coming year?
ATTO designs and manufactures storage connectivity products, and although we don’t manufacture storage, we are a critical part of the storage ecosystem. We regularly work with our customers to find the best solutions to their storage workflow and performance challenges.

ATTO designs products that use a wide variety of storage protocols. SAS, SATA, Fibre Channel, Ethernet and Thunderbolt are all part of our core technology portfolio. We’re starting to see more interest in NVMe solutions. While NVMe has already seen some solid growth as an “inside-the-box” storage solution, scalability, cost and limited management capabilities continue to limit its adoption as an external storage solution.

Data protection is still an important criteria in every data center. We are seeing a shift from traditional hardware RAID and parity RAID to software RAID and parity code implementations. Disk capacity has grown so quickly that it can take days to rebuild a RAID group with hardware controllers. Instead, we see our customers taking advantage of rapidly dropping storage prices and using faster, reliable software RAID implementations with basic HBA hardware.

How has NVMe advanced over the past year?
For inside-the-box storage needs, we have absolutely seen adoption skyrocket. It’s hard to beat the price-to-performance ratio of NVMe drives for system boot, application caching and similar use cases.

ATTO is working independently and with our ecosystem partners to bring those same benefits to shared, networked storage systems. Protocols such as NVMe-oF and FC-NVMe are enabling technologies that are starting to mature, and we see these getting further attention in the coming year.

Do you see NAS overtaking SAN for larger work groups? How about cloud taking on some of what NAS used to do?
We see customers looking for ways to more effectively share storage resources. Acquisition and ongoing support costs, as well as the ability to leverage existing technical skills, seem to be important factors pulling people toward Ethernet-based solutions.
However, there is no free lunch, and these same customers aren’t able to compromise on performance and latency concerns, which are important reasons why they used SANs in the first place. So there’s a lot of uncertainty in the market today. Since we design and market products in both the NAS and SAN spaces, we spend a lot of time talking with our customers about their priorities so that we can help them pick the solutions that best fit their needs.

Masstech, CTO, Mike Palmer

Masstech creates intelligent storage and asset lifecycle management solutions for the media and entertainment industry, focusing on broadcast and video content storage management with IT technologies.

What kind of storage are you offering, and will that be changing in the coming year?
Masstech products are used to manage a combination of any or all of these kinds of storage. Masstech allows content to move without friction across and through all of these technologies, most often using automated workflows and unified interfaces that hide the complexity otherwise required to directly manage content across so many different types of storage.

Are certain storage tiers more suitable for different asset types, workflows, etc.?
One of the benefits of having such a wide range of storage technologies to choose from is that we have the flexibility to match application requirements with the optimum performance characteristics of different storage technologies in each step of the lifecycle. Users now expect that content will automatically move to storage with the optimal combination of speed and price as it progresses through workflow.

In the past, HSM was designed to handle this task for on-prem storage. The challenge is much wider now with the addition of a plethora of storage technologies and services. Rather than moving between just two or three tiers of on-prem storage, content now often needs to flow through a hybrid environment of on-prem and cloud storage, often involving multiple cloud services, each with three or four sub-tiers. Making that happen in a seamless way, both to users and to integrated MAMs and PAMs, is what we do.

What do you see are the big technology trends that can help storage for M&E?
Cloud storage pricing that continues to drop along with the advance of storage density in both spinning disk and solid state. All of these are interrelated and have the general effect of lowering costs for the end user. For those who have specific business requirements that drive on-prem storage, the availability of higher density tape and optical disks is enabling petabytes of very efficient cold storage within less space than contained in a single rack.

How has NVMe advanced over the past year?
In addition to the obvious application of making media available more quickly, the greatest value of NVMe within M&E may be found in enabling faster search of both structured and unstructured metadata associated with media. Yes, we need faster access to media, but in many cases we must first find the media before it can be accessed. NVMe can make that search experience, particularly for large libraries, federated data sets and media lakes, lightning quick.

Do you see NAS overtaking SAN for larger workgroups? How about cloud taking on some of what NAS used to do?
Just as AWS, Azure and Wasabi, among other large players, have replaced many instances of on-prem NAS, so have Box, Dropbox, Google Drive and iCloud replaced many (but not all) of the USB drives gathering dust in the bottom of desk drawers. As NAS is built on top of faster and faster performing technologies, it is also beginning to put additional pressure on SAN – particularly for users who are sensitive to price and the amount of administration required.

Backblaze, Director of Product Marketing, M&E, Skip Levens

Backblaze offers easy-to-use cloud backup, archive and storage services. With over 12 years of experience and more than 800 Petabytes of customer data under management, Backblaze has offers cloud storage to anyone looking to create, distribute and preserve their content forever.

What kind of storage are you offering and will that be changing in the coming year?
At Backblaze, we offer a single class, or tier, of storage where everything’s active and immediately available wherever you need it, and it’s protected better than it would be on spinning disk or RAID systems.

Are certain storage tiers more suitable for different asset types, workflows, etc?
Absolutely. For example, animators need different storage than a team of editors all editing a 4K project at the same time. And keeping your entire content library on your shared storage could get expensive indeed.

We’ve found that users can give up all that unneeded complexity and cost that gets in the way of creating content in two steps:
– Step one is getting off of the “shared storage expansion treadmill” and buying just enough on-site shared storage that fits your team. If you’re delivering a TV show every week and need a SAN, make it just large enough for your work in process and no larger.

– Step two is to get all of your content into active cloud storage. This not only frees up space on your shared storage, but makes all of your content highly protected and highly available at the same time. Since most of your team probably use MAM to find and discover content, the storage that assets actually live on is completely transparent.

Now life gets very simple for creative support teams managing that workflow: your shared storage stays fast and lean, and you can stop paying for storage that doesn’t fit that model. This could include getting rid of LTO, big JBODs or anything with a limited warranty and a maintenance contract.

What do you see are the big technology trends that can help storage for M&E?
For shooters and on-set data wranglers, the new class of ultra-fast flash drives dramatically speeds up collecting massive files with extremely high resolution. Of course, raw content isn’t safe until it’s ingested, so even after moving shots to two sets of external drives or a RAID cart, we’re seeing cloud archive on ingest. Uploading files from a remote location, before you get all the way back to the editing suite, unlocks a lot of speed and collaboration advantages — the content is protected faster, and your ingest tools can start making proxy versions that everyone can start working on, such as grading, commenting, even rough cuts.

We’re also seeing cloud-delivered workflow applications. The days of buying and maintaining a server and storage in your shop to run an application may seem old-fashioned. Especially when that entire experience can now be delivered from the cloud and on-demand.

Iconik, for example, is a complete, personalized deployment of a project collaboration, asset review and management tool – but it lives entirely in the cloud. When you log in, your app springs to life instantly in the cloud, so you only pay for the application when you actually use it. Users just want to get their creative work done and can’t tell it isn’t a traditional asset manager.

How has NVMe advanced over the past year?
NVMe means flash storage can completely ditch legacy storage controllers like the ones on traditional SATA hard drives. When you can fit 2TB of storage on a stick thats only 22 millimeters by 80 millimeters — not much larger than a stick of gum — and it’s 20 times faster than an external spinning hard drive and draws only 3.5V, that’s a game changer for data wrangling and camera cart offload right now.

And that’s on PCIe 3. The PCI Express standard is evolving faster and faster too. PCIe 4 motherboards are starting to come online now, PCIe 5 was finalized in May, and PCIe 6 is already in development. When every generation doubles the available bandwidth that can feed that NVMEe storage, the future is very, very bright for NVMe.

Do you see NAS overtaking SAN for larger workgroups? How about cloud taking on some of what NAS used to do?
For users who work in widely distributed teams, the cloud is absolutely eating NAS. When the solution driving your team’s projects and collaboration is the dashboard and focus of the team — and active cloud storage seamlessly supports all of the content underneath — it no longer needs to be on a NAS.

But for large teams that do fast-paced editing and creation, the answer to “what is the best shared storage for our team” is still usually a SAN, or tightly-coupled, high-performance NAS.

Either way, by moving content and project archives to the cloud, you can keep SAN and NAS costs in check and have a more productive workflow, and more opportunities to use all that content for new projects.