The Growing Interest in Object Storage for Media

After publishing my last blog, I participated in a panel discussion on Sports Asset Management at the Sports Video Group’s (SVG) Summit, “The Business of Sports Production.” While I waited for my afternoon panel session, I attended several of the other sessions in the “Asset Management” track. Each session I attended eventually focused on Object Storage and the promise it holds for media content.

When my session rolled around, there were five panelists in our group all representing different vendors of storage solutions. I fielded the first question, which was what should the audience watch for from NetApp in the way of new technologies for media, to which I answered, “You need to check out our new iteration of the StorageGRID Webscale Object Storage solution.” That one comment seemed to focus the rest of the session, as the audience wanted to know more about what we were doing.

Obviously, the other four vendors on the panel could not be expected to go quietly. They all had their own flavors of object storage solutions, albeit some were licensed solutions from other companies. It was interesting how each one had a different focus, so I thought I would discuss a few of these differences and show you why we think NetApp has the most complete vision in object storage solutions.

Software versus Appliance
As I have pointed out in earlier blogs, most people have an old idea of what object storage really is. They think that it is mainly a way to use cheap commodity disk instead of more expensive storage arrays. They ignore the fact that the old model also required several full copies of the content to protect it. The new model for object storage is one of maintaining content across geographies through a technology called geo-dispersed erasure coding, which requires several data sites and a sophisticated data network to provide data protection and business continuity.

To mask the complexity of this new object storage model, most vendors offer appliances for their customers. This means buying a new storage platform that differs from your existing SAN and NAS storage. When you fill up this new object storage appliance, you simply buy another appliance and hook the units together.

This might work for new environments or smaller operations, but it ignores the fact that most media companies already have petabytes of storage solutions that work well for them today. Can an object storage solution vendor really expect that their customers are going to replace everything they have for one vendor’s object storage solution appliance? Most have fought for years to avoid the “one vendor” solution set.

NetApp StorageGRID Webscale is Software-defined Object Storage. You do not need to buy our disks or disk arrays to get it up and running. The software runs on a standard server or virtual machine and can control disk arrays from a variety of enterprise storage vendors. Don’t get me wrong, we do want you to buy our storage, usually as the first target for the new object storage solution you put in place, but we know that you need to be able to reuse the disks you already have to more easily test and adopt the solution. And with Webscale software-defined object storage, you can do just that!

NetApp has plans to introduce an appliance for StorageGRID Webscale in 2015, but that is for those who only need an appliance. We firmly believe that one size does not fit all and that software-defined object storage is what most media companies will want.

The Interface to Object Storage
NetApp’s StorageGRID Webscale object storage solution interface is S3 and CDMI. Our competitors hype NFS and CIFS interfaces to allow utilization of existing media applications with their object storage. We agree, but only somewhat.

First of all, NFS and CIFS gateways are available today to run with almost any vendor’s object storage. It is not a reason to buy someone’s specific object storage solution. NFS and CIFS utilize metadata from a file system to tell them where the content is to be found on the disk. That means every read request requires a metadata search to access the data. Object Storage works similarly; you request an object using the unique identifier tag and the object storage solution uses its metadata to find the object on the disk. If you add NFS and CIFS on top of an object storage solution, you require TWO metadata look-ups, not one. That adds latency to the asset reads. This is one of the leading reasons that object storage systems are much slower to access data than a standard storage solution.

NetApp believes that object storage is a new construct; that it is best implemented as a new solution. That way, you can get the maximum benefit and maximum performance from this new technology. How do we solve the problem of incompatibility with existing media solutions? Honestly, we believe that problem is going away rapidly anyway. All cloud storage environments are object storage already (you get a unique URL for each asset). Most media applications (such as MAM systems) are becoming cloud aware today, which means that they are becoming object storage aware as well.

One member of the audience at our panel discussion asked how NetApp was planning to get media application software vendors to support these cloud interfaces, I replied that we cannot get these vendors to do anything that their customers have not asked them to do first. This will be like any other change in our industry (think IP video); you the customer need to put pressure on your application vendors to support S3 or CDMI. Think of the storage options that will open up for you when they do!

Information Lifecycle Management and Data Mobility
Lastly, I brought up the Information Lifecycle Management (ILM) and Data Mobility issues I talked about in my last blog. With geo-dispersed erasure coding, our competitors claim that your file is everywhere, or at least accessible everywhere. But, in fact, the file as a full and separate asset is not anywhere. The object storage solution relies on the WAN and the data stored in other sites to read a complete file at any location. This requires significant WAN bandwidth and relies on all of the multiple sites being up and available. While it’s a great protection scheme, the performance characteristics are not going to be a good match for most media workflows.

Without the ability to automatically put the full content asset where it is needed or to store the full content asset on a platform that matches the value of the content, object storage is just another storage interface without much benefit. That is why NetApp focused on adding Information Lifecycle Management capabilities controlled by a policy engine to its StorageGRID Webscale object storage solution. With Webscale, you get the ability to protect data through erasure coding while also having the ability to move entire copies of the content asset data to the location and storage platform that fits your media workflow needs.

Need breaking news content to flow to each and every news bureau in your operation? Just put a metadata tag into the asset telling the system that this content is breaking news and Webscale does the rest (based on what policy you set). What if you want the content to move to a public, private or hybrid cloud environment after a certain length of time? It’s automatic! How about moving old data onto tape? StorageGRID Webscale can do that as well. We doubt that any one appliance will meet the variety of workflow performance demands found in any major media company. With Webscale, you decide what storage platform and performance level (tier 1, tier 2, cloud, tape) is required.

The other vendors on the SVG panel mentioned integrating third party software to provide ILM functionality, but that’s always a difficult proposition. NetApp believes that the tools should natively be a part of the Software-defined Object Storage solution.

How do you find out more about NetApp’s StorageGRID Webscale? We’ll have lots to say about it at NAB 2015 in April (South Hall, Lower Lever, Booth SL13309). Until then, keep watching this blog site for more about the Webscale Story as well as what else we’ll be showing at NAB this year!