October 4, 2017
By Jan Ozer Contributing Editor
The Producer's View

The Changing Face (and Price) of the On-Prem Video Encoder

Five years ago when you purchased an on-prem encoder, it came with its own computer. Today, as more companies invest in general-purpose internal computing in one form or another, that picture has changed. So rather than buying a box, you buy software to install in your internal cloud or container-based computing platform.

Alternatives abound. Bitmovin’s on-prem offering includes “a managed on-premises encoding service on Kubernetes and Docker that works for VoD and Live and offers the same features as our cloud encoding.” You can deploy Encoding.com’s HybridCloud solution “on a variety [of] virtualization platforms (OpenStack, VMware, Joyent), storage configurations (SMB, SWIFT, S3, NFS), and network configurations.” Harmonic’s Electra XVM “is the first product to leverage the broad capabilities of the Harmonic VOS platform and architecture.

Deploying encoders on internal clouds or container clusters makes a lot of sense, as you can very cheaply process your day-to-day encodes on hardware that you already own and easily send any spikes in production out to the cloud.

What’s fascinating is how the merger of internal and cloud encoding is changing the pricing model for on-prem encoding software. Five years ago, when you bought that hardware encoder (or software encoder to install on a single computer), you paid a hefty sum for the privilege and the encoder ran on that single computer. Computer-based pricing clearly doesn’t work when you’re installing the software on multiple CPUs for part-time usage that changes from day to day.

What does work? You have to look to the cloud, though pricing there is also in flux. Five years ago, VOD pricing was based on the minute or the GB, and live transcoding didn’t really exist. Today, this fixed pricing model is changing. For example, Encoding.com now offers Reserved Cloud pricing, where you pay a fixed monthly cost for a cloud instance that you can run at capacity and pump as much encoding through as possible. For customers operating at near full capacity, this is much cheaper than paying by the GB. Hybrik entered the market in 2016 with stepped pricing based on the number of cloud instances that can run simultaneously. You pay them for access to the software, and pay your own hardware costs.

On the live side, some services like Microsoft Azure and Zencoder are still charging by the output hour, a model that Wowza Streaming Cloud undercuts with a $500/month plan, including the costs of the cloud hardware. Or you can buy Wowza Streaming Engine for $1,995, or license it for as low as $65/month and pay your own hardware costs. Softvelum’s Nimble Streamer charges about $80/month for its Transcoder (including all other costs), and you pay hardware costs. These Wowza/Nimble Streamer alternatives are all much cheaper than paying by the hour.

Running on your own hardware should always be cheaper than encoding via a SaaS or PaaS cloud. As the prices for live and VOD cloud alternatives continue to drop, so should the cost of software that you run on your own hardware. Keep this in mind when negotiating pricing for any software you can use internally.

Is the new breed of on-prem software right for you? Looking purely at the economic side, internal processing should be cheaper than a SaaS/ PaaS cloud, though the difference is narrowing. Just 2 or 3 years ago, cloud encoding companies were making obscene margins, making it tough to justify a purely economic decision to encode in the cloud, at least for VOD. Today, with new pricing models squeezing all competitors, encoding in the cloud is truly affordable, narrowing the cost savings of encoding internally.

The bottom line is that you should consider encoding internally, as well as exploring all cloud alternatives. There’s gold in them thar hills in the form of significant cost savings; you just need to find it.

[This article appears in the September 2017 issue of Streaming Media Magazine as "The Changing Face of the On-Prem Encoder."]