US encoder specialist Elemental Technology is aggressively filling out its product line, trying to grow up quickly so it can go head to head with larger businesses like Harmonic – this week adding a statistical multiplexing product and improving its existing video operations management module Conductor.
Elemental first came to our attention when Faultline wrote a parody of the CEO of rival software encoder specialist, Envivio. Elemental’s CEO Sam Blackman thought it was hilarious – and clearly has no plans to make the same mistakes that Envivio has. But there are parallels between the two companies – as Elemental slides sideways into neighboring markets – Origin servers and statistical multiplexing (stat mux), Envivio is also trying to augment its product line, but has chosen Ad insertion (Halo) and Cloud DVR, as well as just-in-time packaging.
For Elemental, this is barely 6 months into pushing Elemental Delta, its new origin server that also handles just-in-time packaging for adaptive bit rate (ABR) technologies – but here it is extending its product line once again.
The truth is that the encoding market has changed beyond all recognition. It used to be that around $3 billion a year changed hands, mostly on specialist “real time” encoding, and far less on VoD compression and preparation. Since ABR schemes were invented, virtually no-one has focused on replacing real time encoders, so Ericsson and Harmonic have had to diversify or live with simple replacement and support revenues.
At the same time, more and more companies have come to understand codecs and make a living out of them. In the new ecosystem, the open source versions of X.264 and X.265 have placed a ceiling over encoding pricing. Now withtwo programmers and an Intel Xeon server, you can build an encoding platform of sorts.
A few years ago the top five players, considered to be Arris, Cisco, Ericsson, Harmonic and Thomson Video Networks, accounted for 70% of the encoder market. For the most part, their products were based around either FPGAs or ASICs which needed years to develop – in particular the camera chips from Ambarella. Although the price per new real time channel did not fall too rapidly because of the high cost of entry, it is falling now.
The three companies that have challenged this are Envivio, which to be fair blazed a trail in software only encoders, followed by Ateme, which came at this market from the contribution encoder market that it had been in for years, and Elemental – a start-up which took its lead from harnessing the native power of an Nvidia GPU.
But the very act of putting encoders on general purpose servers means that the market will commoditize. Alongside the open source encoders, this lower price of entry into this market has opened the floodgates. Only the sudden need to get to HEVC first and then on to UHD encoders has kept pricing up in this market.
Cumulatively, server power has increased and also dropped in price thanks to cloud arrays, and GPU and other video acceleration elements have emerged in servers. This shift has taken away the benefits of creating specialized encoding platforms – that has been left to chip makers who make to put an encoder in a phone, rather than in the cloud (see separate article this week on Imagination).
The other thing driving the encoder markets, and not likely to sustain pricing, is the emergence of the Asia Pacific markets in pay TV. Here, legacy product lines are of no help, as there is no need to do MPEG2 real time encoding, because most of the new operators in these markets want the savings of H.264 day one, and HEVC as soon as possible. This shift in demand isn’t because these operators are especially new. Instead, it’s because they use digital distribution to begin with and not legacy analog – wanting more channels for an enlivened and enlarged middle class that appreciates the value found in digital content and quality.
So while the broadening of the big five product lines has happened over a seriously long time, Elemental has a creeping feeling that it must copy this shift and do it fast – partially to drive revenues, but also as an antidote to price dilution, and to have more irons on the fire.
What happens is simple explains Keith Wymbs, head of marketing for Elemental: “Customers of our ABR Elemental Server (for files) or Live, ask if we can help then with their legacy video streams. Can you do from MPEG2 to the set top? Stat mux is all about the telco market wanting UDP in an MPEG transport stream, or for multicast and cable companies”
“Ten years ago, one channel equaled one box. Today in ABR, the density is far higher and the ratio is many to many (channels to boxes). Stat mux is particular important for delivery out to QAM or to a satellite uplink, but also for moving video around internally, within the network,” Wymbs said.
“But if we want to do that we have to adapt our management platform conductor, so that it can handle thousands of streams simultaneously. It needs bulk configuration management and scale.”
The truth of the matter is that as prices fall in encoding, operator management want to forget about it and there are now multiple ways of doing that. The first way is that they leave everything to one vendor, who is their first port of call for all new capabilities. The second approach is to put it in all in the cloud and the third is to have it management and hosted by an outside company, within the cloud.
Sometimes to get the best of both worlds an operator wants to deal with one supplier, but experiment with data innovation at the codec level – which is why Ericsson has a virtualization system for cloud encoding, and Harmonic something similar, so that it can use Envivio and Elemental elements, but wrap them in an established Ericsson or Harmonic Workflow.
And for Cisco, we have seen deals where it takes a license to a rival’s processes, and includes them in its product line, but the customer is blissfully unaware that it is buying an Elemental encoder, with a Cisco badge on it.
While that may seem good for business, we know that these types of setups leave all the caring and looking after revenues for the incumbent. While the fresh innovators at the codec level get some money, it is the incumbent who takes the headache away, and gets more of the cash. And for statistical multiplexing it is mostly about tight integration with all the steps in the route.
For instance, if you build a stat mux, you may have ten TV channels with their I-Frames all issued at the same time – which leads to a very “lumpy” flow of data and the opportunity to use up too much bandwidth. So by going back to the encoder and asking for perhaps a few more frames in the last group of pictures in a variable bit rate stream, the system can fit in more channels for a given bandwidth. What is needed is this feedback loop to the encoding element, and a second chance to re-encode one or more of the TV channels at a point in time, slightly differently. And that needs tight integration between stat mux and encoder and they have to sit near each other in a workflow, side by side in a cloud, all working in real time, but working faster than real time to get the second chances.
Systems like this need to be able to do things like multiplex streams from different codecs (HEVC with H.264) some in 4K and some in full UHD or plain HD. The goal is to make sure that the bitstream fits down the delivery pipe without touching the edges and negatively affecting performance.
In the past, stat muxing gave a gain of up to 40% in bandwidth, but Wymbs says, “Encoders are more efficient these days and we have seen a gain of around 25% to 30%.”
Interestingly, as we commented last week, Envivio just won an award from Frost and Sullivan for having the largest revenue in the Live Video Transcoding Market. We ridiculed this last week, because it came out a day before the Envivio results, and we asked if the illegal step of allowing Frost and Sullivan access to its forthcoming numbers was behind this.
The numbers duly came out and revenues leaped from $9 million in the immediate past quarter to $12.7 million, capping the previous year’s equivalent quarter of $12.5 million – in other words Envivio is back on track for $50 million a year revenue levels, something that it said just prior to it going public, whereupon its revenues instantly collapsed. Now it would have us believe that it is back on the road to recovery. Revenue for fiscal year 2015 was $41.6 million, compared with $43.2 million in fiscal 2014. It lost just $1.6 million for the quarter, but $15.6 million for the year.
The company initially filed for an IPO in 2011, finally went public in 2012, and put upwards of $75 million in the bank. Today it still has $38 million, and had $48 million a year ago. Its CEO said at the results that “Envivio is experiencing particularly strong wins in use cases where it can offer an all-in-one or ‘converged’ head end, such as IPTV/multi-screen.”
The company picked up deals at Bouygues Telecom in France to do multiscreen in the cloud and signed Taiwan operator Chunghwa to use Envivio for its 4K HEVC trial. It says it has significant traction in HEVC, but this is just an example of what we have always said – in encoders, once you switch a system on, you want to work with the same company and mostly just forget about it. Chunghwa could have had 4K about 18 months before, if it had switched to Elemental – but it looks like it was quite happy to wait until it came out of its existing supplier.
Elemental Technologies will launch its new services at NAB in the US this week, demonstrating its software-defined video system with statistical multiplexing, end-to-end redundancy and multiscreen delivery. Naturally the stat mux is also software based, and as we said, multi-codec and multi-resolution.
And Elemental rolled out its first customer – Latin American pay TV operator Telecable in Uruguay, where it only claimed 15% to 20% bandwidth savings. TCC is part of Mexican media company Televisa, one of the Latin American video giants, after such companies as America Movile, DirecTV and Telefonica.
As the industry shifts, there will soon be no encoder companies left – they will all either grow sideways to be end to end multiscreen companies or they will go under and be assimilated.