Cloud, On-Prem, or
Something in Between?
The question of whether to rely on cloud infrastructure or to invest in
hardware is an important one for the video industry. Not so long ago
cloud was the future of just about everything. Flexible, scalable, and
dynamic; cloud computing handed development teams virtually infinite
resources for developing, testing, and building powerful global solutions
without the overheads of building and maintaining on-prem hardware
infrastructures. Lately, however, we have seen this sentiment changing,
as many companies are doing the math on their cloud expenses, and in
some cases coming to the conclusion that on-prem is a better option.
WHY CLOUD COMPUTING MAKES SENSE FOR VIDEO
There are a number of very strong arguments to move your encoding platform to the
cloud. Some revolve around the limitations and overhead associated with traditional
hardware encoders, and others are based on the scalable nature of cloud infrastructures.
Scalability and Performance
One of the huge advantages of cloud computing is the potential to instantly scale your
processing capacity as demand requires. A containerized encoding system can realize this
potential by allowing video segments to be encoded simultaneously across a number of
cloud instances. When we combine the scalability of the cloud with containerized video
encoding, we have the ability to complete large batch encoding jobs in a fraction of the
time. In fact the time taken to complete a batch job becomes more of a budgetary decision
than a processing limitation; i.e., the speed of any given encoding process is only limited by
the number of cloud instances that you are willing to pay for.
For VOD and OT T providers that add new content to their libraries on a regular basis, the
cloud solution performs much more efficiently. On one hand, they can ramp up their capacity
when new content arrives, getting the new content online faster. And on the other hand, they
can scale back to zero when there is nothing to encode. Live streaming also benefits from the
same efficiency. Instances can fire up as an event starts and shut down as soon as it ends.