Every few years (or months) it is mooted that “the protocol wars are over,” with the hypothesis that even the most basic of protocols have surpassed what was possible five to 10 years ago, bandwidth is cheaper, 4G/5G will reduce the need for economy on bandwidth, and so on.
However, protocol development and advances seem to be accelerating, driven by new use cases and new dynamics between vendors. It’s all quite exciting, so we thought we’d catch up on recent developments and what this might mean.
What has been going on with Teradici—both technically and terms of their business model—is really interesting, and represents a key data point in where the protocol ecosystem may be heading. Teradici’s flagship PCoIP protocol was designed to leverage their PCoIP hardware products, and was promoted by VMware as their main VDI protocol for years. What VMware uses is essentially a rather specific version of PCoIP, which has very different capabilities compared to the variety of PCoIP used in Teradici’s own Cloud Access Software products (beyond graphics and decompression/compression). It makes sense for VMware to focus on their own in-house Blast Extreme protocol, as cutting a third-party protocol cost out lowers the cost of the solution.
Although VMware still officially fully supports and endorses PCoIP, I’m not convinced their heart is truly there, with some bugs fixed in Blast Extreme but not in PCoIP.
Teradici’s own variety of PCoIP is very different from the integration VMware is using, and is included in Teradici Cloud Access Software. Like Citrix HDX, Teradici has focused on a multi-codec hybrid protocol strategy, meaning the type of data on a screen is detected and compressed using the most efficient codec. (For video, this is often H.264, but for text or CAD line data it’s often others; different regions of a screen can be transmitted via different codecs.) Like Citrix, Teradici also offers options for full-screen H.264 and various other vanilla codecs.
Teradici’s business model
With the revenues from the VMware variety of PCoIP potentially drying up, and VMware promoting their own Blast Extreme, Teradici has had to adapt. Rather smartly, they focused on the cloud, SaaS, and DaaS markets, where there are an awful lot of vendors who don’t want full VDI solutions from VMware/Citrix, and instead just want a protocol (and sometimes a broker) to deliver pixels efficiently. Additionally, many DaaS/SaaS vendors want the ability to move about Google Cloud, AWS, and Azure in order to follow pricing. This in turn is driving innovation in the broker vendor space, as Jack wrote recently about Leostream.
While there are cheaper out-of-the-box options that are good enough for some users, Teradici allows software providers and clouds to deliver a high-end protocol for discerning users, a use case exemplified when VFX vendor Foundry announcing their intention to partner with Teradici.
Teradici’s technology advances
Teradici has suffered some misleading marketing and FUD around their capabilities. It’s not uncommon to see PCoIP benchmarked against systems enabled with other vendors’ GPUs, while the PCoIP hardware that the protocol was optimized for is left out of the equation, particularly on the end-points. It’s also very common to see the “PCoIP” used in the comparison to be VMware’s implementation, rather than the current core Teradici offering. In my opinion, their technology is some of the most innovative and exciting in the protocol game, but often it is evaluated in configurations where it can’t show its full capabilities.
In recent years, Citrix has been consolidating their protocols and sorting out their defaults. Originally HDX 3D Pro was a premium SKU, but customers aren’t terribly keen on paying separately when they’re also paying for all the other VDI stuff, too. I think we’ll see less focus on the HDX brand (especially HDX 3D Pro) as the marketing shifts to EDT (Enlightened Data Transport), which is their term for rationalizing everything together.
Framehawk has essentially been replaced by UDP options, and the defaults have moved to selective H.264 as well as more automated technologies, which can adapt to network conditions. Selective H.264 is very similar to Teradici’s hybrid approach, with different regions transmitted by the most suitable codec. A good overview is available from Citrix.
Citrix added support for H.265 (the successor to H.264) about six months ago. While indicating leadership in the protocol wars, it’s probably not a game changer, as most legacy endpoints out there are not H.265 hardware decode enabled, and it’s unavailable on server OSes. Therefore, the selective encode hybrid protocol remains H.264.
Last month, Citrix’s Muhammad Dawood blogged about the HDX protocol advances in XD/XA 7.18. The latest developments focus on progressive display and interruption, building up the screen in quality as needed. There’s no need to go pixel-perfect immediately on a rapidly changing screen, and it makes sense to abort screen construction if user has started doing something else. These kinds of innovations are particularly useful for scenarios including AR and VR, as well as mainstream VDI.
VMware has been focusing on their Blast Extreme protocol. When first released in 2016, it was initially a rather basic full-screen H.264 4:2:0 implementation. H.264 4:2:0 is just a bit fuzzy for most regular VDI users, let alone for those fussy about visual quality, so VMware has recently added H.264 4:4:4 full screen support, which has better quality. This comes at the expanse of bandwidth, but at least it matches the full-screen H.264 offerings Teradici and Citrix HDX 3D Pro have offered for many years. Many in the industry are of the opinion that VMware’s H.264 implementation is very well done and, while it should be pretty equivalent, it actually outperforms many of its competitors.
One issue is that pure H.264 has limitations for integration with products such as NetScaler and SD-WAN, which offer de-duplication and caching over large enterprise app or VDI farms. This means that VMware probably needs to look to expand their protocol technologies significantly to win over existing large Citrix customers who care about bandwidth.
There’s a continual battle for VMware and Citrix to grab each other’s existing user base, while also fighting off the nibbling effect of lighter and cheaper VDI solutions and alternatives. So, protocol innovation, efficiency, and performance remain key factors to retain customers. If changing to an alternative means more hardware or networking (bandwidth), it’s a big obstacle to a switch.
The competition from ‘good-enough’ alternatives
RDP/RDS is now very good. Configured out-of-the-box for high image quality, it’s not the most efficient on bandwidth, but there are plenty of organizations that have the bandwidth available and can make the costs work. Low cost VNC solutions for customer support use cases work fine.
With users reluctant to spend on additional licensing, VDI and cloud vendors are throwing in more for free. If you can cut a customer’s costs by avoiding third parties, they are generally more content with paying for your product.
Do it yourself (DIY)!
As VMware found, using a third-party protocol (Teradici) for a long time raised the final cost of their solution, so taking it in-house with H.264 Blast Extreme made sense.
H.264 is a natural choice of a standard, ubiquitous technology to develop your own protocol on. It’s a path many cloud and SaaS providers follow, as it’s good enough for many situations. Doing it well and maintaining a protocol is expensive and risky, though—if your protocol is ropey, your entire product experience is.
Of the big cloud providers, AWS’s NICE DCV acquisition is a perfect data point of acquiring a ready-made protocol and team. Now thrown in for free on AWS instances, AWS can offer a high-performance Linux and Windows VM protocol, adding value to the “why AWS?” choice.
We can see there is a lot of innovation and development going on. Most vendors are actually following a fairly similar path, with many consolidating on some sort of H.264 base. And overall, cloud, DaaS and SaaS are providing opportunities for vendors beyond VDI.
In part 2, I cover some innovative protocol use cases, and the future technological developments they may drive—I may even pass some opinion on who is winning the protocol wars!