Creating a Din in the Cloud Market

Recently I spoke with Mike Chase of dinCloud regarding its desktop virtualization offering. I hadn’t actually come across dinCloud before, as it is a fairly youthful company. However, its VDI offering—which it refers to as HVD (hosted virtual desktop), making an important distinction, because the solution can be on-premises, off-premises, or hybrid—is effectively a separate, purpose-built VDI infrastructure that can be deployed as public or private cloud.

dincloud

dinCloud isn’t based around technology from any of the “big boys” in this arena—they eschewed the offerings of VMware and Citrix on the basis that they believed, among other things, there wasn’t a great industry collaboration from either of them. Mike also offers the opinion, which I agree with, that Azure and AWS are primarily focused on servers, rather than desktops—although Amazon is now offering WorkSpaces. dinCloud’s primary focus was on creating the right kind of appreciation of the hardware environment that was required for users to run desktops, concentrating on creating a solution combining ultra low-cost hardware with low-latency networking and super-fast disks.

The final solution it brought together is a custom solution called dinStack, which is composed of technology from fourteen different providers. Hardware is based around Supermicro custom servers such as those used by Netflix, Juniper, and Nutanix, using 100% AMD chips to ensure that memory runs at full speed; also provided is object-oriented storage using the S3 protocol, similar to that used by Amazon, FaceBook, and Microsoft. InfiniBand is also a key piece of this solution, providing lower latency than Ethernet, virtualizing all network connectivity, and avoiding the requirement for a hypervisor restart when provisioning I/O. The hardware runs a KVM-based hypervisor, but dinCloud is licensed for Hyper-V as well as KVM, providing the best of both licensing worlds for MS and Linux users.

Mike reports that when testing industry-standard thin clients for compatibility with its solution, dinCloud took the unusual step of using YouTube video streaming as its primary benchmark. Many of the thin clients utilized didn’t manage to pass this test, and it found that the ideal “thin clients” would in fact be ultra low-spec PCs such as those with Atom processors, or the likes of Chromebooks. But the focus on streaming video as the primary client compatibility test should indicate that the dinCloud solution has been designed primarily with the desktop user in mind.

This is not to say that dinCloud can’t virtualize servers as well—it can. Machine resources can scale up to 128CPUs, 2TB RAM and 5TB of storage (even on a desktop, should you wish to!) When accessing the dinManage web portal, you can spin up virtual servers or desktops in a virtual private data center through a dedicated virtual firewall. A point-to-point VPN (virtual private network) can then be configured, allowing you to put in systems such as Active Directory domain controllers and then replicate them. Unlimited traffic is allowed over the VPN, making hosting servers in there a much more palatable offering.

This brings me to one of the major advantages that dinCloud has over the competitors—price plan. Both AWS and Azure suffer from what I’ve previously referred to as the “Ryanair” factor—the constant assessment of unexpected extra charges, and the difficulty therein to predict how much a virtualized solution will cost. dinCloud assesses licensing on a per-customer basis—it has En Pointe alongside it—and then produces a cost that may possibly be cheaper based on your existing licensing. However, the base rate is set at $20/desktop for consumers and $40/desktop for business, increased slightly if you want to run full Windows 7 (although this seems to be 32-bit only) or Windows 8 rather than a 2008 R2 session. There are no hidden charges whatsoever—it’s a flat rate, all-you-can-eat plan—although there are minimum seat purchases for several of their packages, mainly the business ones. The additional charge for storage is 15c/GB each month (40GB is included in the base price), with no transfer fees. Additionally, the En Pointe partnership also means they can guarantee MS licensing compliance, which is something I haven’t yet seen another cloud provider offer as of the date of this article.

dinCloud also appears to have other advantages over competitors, such as the ability to change resources (such as disk sizes) on the fly, dependent on the underlying OS, unlike Amazon, which requires detachment of disks to do this. VoIP is supported as standard, as is Netflix (on its webHVD offering), which is something no other cloud desktop provider can do. Printing is done via Google Cloud Print, which requires a Windows driver download and registration of your printers. Once that is done, printers can be not only used but shared with other users, giving you the ability to print things to printers in remote locations for other users to access. All traffic in and out of the virtual datacenter is run through the ThreatStop reputation filter, again without any extra charge.

It’s interesting to note the collaboration dinCloud has with Google, especially since Google and VMware (which together own Desktone, one of dinCloud’s competitors) have been publicly high-fiving each other over DaaS quite recently. Google appears to be collaborating with dinCloud to use the virtual desktop solution as a “bridge” for Chromebook users who are accustomed to using Windows exclusively. There’s also integration with HTML5 for accessing the solution without a client via Google Chrome, and this will be extended to other browsers in the future.

My reservations about their name aside—it isn’t the catchiest—dinCloud looks like quite a feature-rich solution, but the main selling point for me is the flat-rate charges. Many cloud projects struggle to put a definite cost on the hosting of desktops and servers, but dinCloud’s flat monthly fee sets it apart with regard to this issue. There’s always the chance that as more and more people adopt hosted solutions, ISPs may decide to cap data access and introduce more charges that way, but that’s something that dinCloud itself will have no control over.

Can dinCloud compete with the big boys of this arena? Given that they have produced a solution that looks more like the idea of ITaaS than anything else out there, it’s possible that it could. The proof, as ever, will be firmly in the pudding, but at first glance it looks like it is a very enticing addition to the desktop virtualization melting pot.