INSTALLING MEDIA SERVER:
BANDWIDTH vs CAPACITY
In the last post we talked about how you can estimate how much capacity is needed for each production and how that can be used to estimate total capacity for a network storage solution. This post will compare capacity and bandwidth for video network solutions.
Users are familiar with the speed and convenience of directly attached storage when working with media assets. Unfortunately, this type of storage isn‘t suitable to collaborative workflows, as projects and assets can‘t be shared conveniently between different workstations in real-time. Network storage solutions are designed specifically to share media files across all platforms and application while providing consistent high performance for all workstations.
You can have either a complex or simple calculation to determine the capacity you would need for online storage. Video storage is defined by either Online or Offline...next blog post...but for now lets use the online model. Assume you need storage for 50TB of 4K ProRes online material. Most online storage systems use spinning platter HDDs of varying capacities, the most common now are 1,2,4,8,10TB drives. Storage drives are getting cheaper but large capacity - high performance still cost expensive blah blah blah.
Especially in the post production world, you pay more for storage that can scale. Hardware vendors will claim that their solutions are scalable, but does it scale internally or externally? Does it scale by adding more drives? Or does it expand externally by purchasing an additional unit with separate drives and connecting that into an existing ecosystem. Some solutions have scaling methods that are designed seamlessly (plug and play), while others may necessitate migrating your entire media library off and back on again to redistribute your media across the new drives/new units. The scaling applies not only to capacity but also to how much bandwidth is available. Some hardware scales capacity but not bandwidth.
QFHD (3840x2160) ProRes HQ according to Apple's ProRes White paper doc estimates 707Mb/s or 318GB/h for 24fps footage. Thats 707 megabits per second and 318 Gigabytes per hour. These numbers are your bandwidth and capacity numbers respectively. In order to have one stream of video across your network to your workstation you'd need an infrastructure that could support transferring at least 700 megabits per second through every point in the network chain.
That means your server, switch, cables, and workstation connections(NICs) all must support at least 1 Gigabit connections. That is only one layer of video, and thats assuming no other traffic on the network for that to play back smoothly. If you have two streams (ie. two camera shoot) you need to double the bandwidth needs to 1400 megabits per second meaning that a single one gigabit connection would not be sufficient to stream those files across your network.* A dual 1 Gb connection may be sufficient however dual 1 Gbit connections are never truly equal to a 2 Gbit connection.
Remember that these requirements are for 1 and 2 streams of content from the server with no other network traffic. If you had a multicam sequence with 5 cameras pulling 4K from the server would require 3535Mb/s and at least a 5Gb/s and it would make sense to move to a 10Gb/s connection for that amount of traffic. And that is for just one workstation. Three workstations pulling two streams of ProRes 422HQ QFHD content simultaneously would equal a load of 4242Mb/s.
IMPORTANT NOTE: Decreasing the resolution of your sequence or the preview window does not affect the amount of data being pulled. A computer will still pull the entire file across the network regardless of your project settings.