In a series of posts we've been outlining the general challenges with scalability of different types of software within large enterprises. In this episode, we look at the scalability of Media Asset Management platforms.
We’re all familiar with Moore’s Law that says hardware performance doubles every 18 months to 2 years. A similar trend is discernible in storage capacity as well. But technologists at large media organizations will tell you that while the power of hardware is increasing “only exponentially” the growth curve of video content is even steeper. (I might call it “explosive.”)
Scaling gracefully, is needless to say, a core requirement for an enterprise media asset management initiative. For example, several steps in the digital supply chain -- transcoding and conversions, file management, search indexing -- are processing intensive. What, then are the common challenges customers face in their quest for scale?
First and foremost, consider the architecture of the MAM software itself. Vendors may say their products are nothing but shining examples of Service Oriented Architecture but not all SOAs are created equal. Do the APIs perform efficiently? Are the interfaces to other systems well defined and documented? How well does it consume services from other systems? Usually, when vendors say SOA they’re talking about exposing an API but they don’t refer to being able to gracefully consume services from other players in the digital supply chain.
Predicting storage and processing requirements is a tough task, but recognize that not all content is equal. Have you provisioned adequate capacity for both the assets themselves and for the creation of search indexes? Do you archive or purge content proactively using business rules and usage patterns, or do you wait till the disks get full and then look for a solution? Some MAM solutions may offer “garbage collection” type of services to delete unused content that’s taking up space.
Based on your capacity planning for the present and future, does the product let you add “services” as content volume grows? Can these be distributed, i.e., can each service run on its own server? Can you distribute a service over multiple machines? These are the questions that usually come up in scaling up MAM projects.
In addition, for multi-location operations (e.g. with East Coast and West Coast teams), do you need specialist file transfer software and WAN accelerators? How well does your MAM software integrate with this? Don't overlook these aspects at the beginning and revisit the assumptions at least annually and make any necessary tweaks.
Giga, Tera, Peta easily roll off our tongues now, and we will likely keep updating our Greek vocabulary going forward. Depending on the volume of assets, you’ll need a plan for content storage. This could involve a hierarchical storage system with online, near-line, and offline tiers, and you’ll need to make trade offs and balance storage costs vs. speed of access to content. Automating the movement of files between different categories of storage based on business rules is also part of the solution.
Search performance will very much shape users' perception of a MAM system implementation effectiveness. Video search is a resource intensive process. Developing a good metadata model and putting in place a content pruning plan based on asset usage analysis will help, but you’ll need to deploy both smart and brute force.
Digital Supply Chain Bottlenecks
If the digital supply chain is not properly designed, bottlenecks in one part of the chain can cause pile-ups in the other parts. Bottlenecks can be human or computer related. For instance, you may experience queues in the ingest service, while the conversion service is idling and operators may be waiting for processing to finish. This too will require an initial framework that needs to be regularly updated with the evolution of your business.
Successful MAM projects that have scaled well tend to pay attention throughout the project lifecycle, from the initial product selection to configuration design, to ongoing management and optimization.
For a detailed evaluation of how the major MAM vendors stack up against these requirements, consult our exclusive research.