Cloud Field Day: Droplet Computing – Any App, Any Where, Any Device
Yes, we have heard that message before from many, many companies in the past – but how many companies have REALLY delivered on that promise. And I mean REALLY delivered on that promise…
The trouble has been that in order to achieve the laudable goal in the past you have needed a truckload of infrastructure to deploy either a server-based or virtual-desktop-based environment – and run those Apps from an expensive data center location usually in the context of a Windows Operating System. Once I said at a user group in the US that putting desktops in the data center (the most expensive computing environment known to human or beast) is unlikely to result in massive cost savings. As I recall, one person stood up and clapped me for my honesty. It doesn’t matter how many TCO calculators you throw at the approach, the data center business is an expensive one. And one that according to our public cloud vendors, is one business customers can’t wait to get out of.
The Technology – Droplet Computing
At the Cloud Field Day in Silicon Valley back in April – Droplet Computing came out “stealth” to reveal they have developed a client side container technology, that will allow practically any application to run in the context of web-browser. Here’s a quick overview of that container technology:
So, there’s a couple of components going on here. But in simple terms, the container sits inside a web-browser which essentially offers the runtime environment, surrounded by a series of supporting “libraries”. Possibly the most significant is the use of WebAssembly which handles the job of intercepting the machine code generated by the app sitting in the container. A clunky analogy would have the web-browser like the virtualisation layer, and the container is like a VM, with the user app sitting inside the VM. All this is done however, without the bloatware of either a server-side hypervisor or client-side virtualisation and an “operating system” sitting inside that VM. That was tried in the past with technologies such as VM Player, ACE, Workstation or Fusion – or worse still downloading an entire VM from the corporate data center – just to run a measly little Windows App. Now I’m not saying that VMware Horizon or Citrix XenApp are “wrong”, it’s just for many they were sledgehammer technologies trying to crack a nut. Of course there the physical system that end-user sits on needs operating system – but that could be almost anything you care to think of.
So, this is ultra-compatible. How compatible? You could easily “natively” run a Windows based App inside a Droplet Container under an Intel-based Apple Mac. All without installing Windows or having to power-up or resume a Windows VM. In fact, the processor wouldn’t even have to be Intel-based, it could be an ARM based processor if necessary. This opens the door to being able to those Windows Apps running on chipsets for which they were never designed.
That lack of Windows requirement stems from the use of “wine” from the world of Linux. Wine has had a sketchy history in the world of Linux as a way of natively running Windows Apps under a Linux context – but it has moved on and improved over the years.
In terms of the underlying physical system – all that’s really needed is a relatively modern web-browser with WebAssembley enabled – which is sufficient to support the Droplet Container. Currently that’s Chrome v6.0, Firefox 5.2, Safari v11, and Internet Explorer v16. Mobile device such as Android-based phones/tablets (Android v6.2) and Apple iPhone/iPAD (iOS v11.1).
So, in a nutshell. Droplet Computing does for desktops what container technology has done for server-side code development and code distribution. Adding a layer without the overhead of a virtualization layer+operating system and other dependencies.
The Use Cases
ANY, ANY, ANY means MANY, MANY MANY.
But let’s start from scratch. The end-user computing world is a very different one from the narrow world of Windows PCs, and the old days of “Virtual Desktop Initiatives”. Whilst it would be foolish to discount server-based computing and VDI, neither succeeded in going mainstream – or become the de-facto way that users got their apps. They remained corralled into a particular niche for Dilbert-style users who sat in their cube all day. That way of working is on the decline with many of us being more mobile or working-from-home (WFM or should that be WTF?) – plus we all now have at least three devices (if not more) in the shape of laptop, tablet, and smartphone.
Whilst attempts have made to duplicate apps across those device types – this has resulted in compromises either in the “app” driven world of the iPAD or in the web-driven world of Office365. It’s always meant some uneasy compromise of the experience – which support folks have to excuse or explain away. So, this new way of working has spawned attempts to bring everything under one house via things like VMware’s WorkspaceONE – a portal to a plethora of different ways off delivering apps (SaaS, Virtual Desktops, Application Packaging – like ThinApp, AppVolumes, Microsoft App-V and so on, and on and on…). For me the difference with Droplet Computing is they are offering a net-new method of delivering the App – in a container, executing on a device of the end-users choosing with their preferred web-browser. Incidentally, I still think these one-stop-shop “App Stores” will still be needed for ID management, entitlement and security reasons – but I can see Droplet Computing being the Apps that are advertised there – to be downloaded and run on the end-users’ device. And of course, if you must have centralised VDI they are possible target too…
Clearly, legacy apps will be an important market – but I personally believe that this approach will pay dividends for new applications as well as older ones. Although to be fair it’s those older applications that often prove to the bane of everyone’s life. Often they’ve been developed in an OS with lower security requirements – and that often means ‘breaking’ the rules and regulations about OS hardening – just to make them work. The older they become the more they are likely to break as their dependencies themselves become incompatible or discontinued. This then has a knock-on effect to other important requirements such as meeting compliance during an external audit, or merely ensuring the apps are as quick and reliable as they once were. The dizzying releases of Windows and their associated Apps means it’s really impossible for an enterprise to freeze its world based on a particular approved “build” and blend of OS/Apps. This just doesn’t sit well in a BOYD era where CorpIT has no clue or control over the end-point the user chooses – never mind that that they may be using a form-factor such as tablet. So, Droplet Computing’s container vision and technology offers a tantalising promise of escaping these limitations and restrictions.
There are some interesting parallels between the early days of virtualisation and what Droplet Computing is doing.
Firstly, there’s a low-hanging fruit market of legacy apps that are still used by business but won’t be supported or won’t work on new operating systems. That includes App’s developed by ISVs who may not actually be trading anymore. The cost of rewriting those legacy apps far outweighs the usefulness to the business, so away of extending their life time beyond meaningful usage is appealing.
Secondly, Although the software running in the container is unmodified and runs natively – customers should be aware that Droplet Computing is not responsible for the licensing policy or support agreement of the ISV. Of course, if they have ceased to operate there’s a great deal of leeway there, but if the ISV is current – they might decide (as they did with 1st Gen virtualisation) simply not to support it or licensing it in a such away as to reduce its competitive value. For instance, a Droplet Computing user license allows the end-user to run 3 copies of the Droplet Computing software – enough to cover the 3 most popular devices a user might use (laptop, smartphone, and tablet).
However, the ISV might choose to charge the business 3 times for an application that has been “installed” on three different devices. Remember many of these legacy applications that represent the low-hanging fruit have quite antiquated licensing policies that are often the bane of many Citrix XenApp or Horizon View admin. That said, the potential cost savings from not having to run the older infrastructure – and having that app execute on expensive compute (the data center) but on cheap compute (the laptop) – could outweigh that restriction. Put simply it might be cheaper to suck up the additional license costs, to save money elsewhere. Personally, my hope (the same hope I had with Gen1 Virtualisation) is that ISVs review their licensing policies with a view that anything that drives consumption also preserves market share – and that it’s not in their interests to corral their horses and carriages in cycle in order to “protect their revenue stream”.
Looking back on this paragraph, I’m perhaps over-egging the impact of these licensing considerations. Perhaps ISVs have woken up the multi-device world we now reside in, and these antiquity licensing policies are a thing of the past?
So, its early days for Droplet Computing. They have secured their first round of VC funding and come out of stealth – and they are on the cusp of releasing their first release candidate 1.0 GA. I hope to get some stick time with the technology, as I believe getting one’s hands dirty is the first step to learning the advantages, disadvantages and limits of technology. I’ve waited a while for a truly new and innovating technology to grab my eye. And not just a rehash of existing bits and bytes. I think what Droplet Computing is doing is very, very interesting – and they are certainly a company to keep on your radar.