It is said that VDI as a concept is straightforward and a compelling proposition. Centralise your services to reduce the desktop management complexity and enable a more cost-effective method of updating desktops. In addition to this, the option to support branch office/remote users from a centralised location can also allow you to reassess network link costs between sites, and indeed VDI allows you to deliver greater productivity through “anywhere, any time” universal access.
In centralising desktop services there’s obviously a requirement to understand the performance and operation of your applications in a virtual host – but, proving that the VDI implementation works technically will be for nothing if you have users who have a poor experience of using the shiny new VDI solution. Indeed, with the rise of mobility, either on mobile devices such as Apple’s iPad or on Microsoft Windows netbooks connected through 3G cards, companies will increasingly rely on networks that experience higher latency and packet loss than on a LAN.
If your organisation has remote users – consider that the impact of centralisation on their desktop experience can be very different: and not always in a happy way.
Notes from the Coalface
In a recent blog post David Freund, from EMC, detailed some of the issues that were encountered on their Journey to a Private Cloud. Like many other organisations, EMC chose a phased approach to introducing VDI – allowing them to learn about the technology and make best use of it. Feedback from their Proof of Concept was positive, which is a similar experience with many other PoC’s. EMC users compared their VDI experience with their desktop delivered from their physical desktop. It was found that just over half the POC users would use VDI for a secondary desktop, but only a small minority of those said they’d pick a virtual desktop to replace their current machine.
Obviously, while VDI may reduce desktop management complexity its not an effective solution if the service it delivers to users is perceived to be much more cumbersome and difficult to use. In EMC’s case, users reported slower performance when printing locally and accessing local drives—especially for those users in Europe and Asia. EMC’s lab testing uncovered a significant performance bottleneck in Microsoft Windows’ Remote Desktop Protocol (RDP) on links with high latency.
The latency “pain threshold” in their case ended up being around 100 milliseconds. For local users, where network latencies are typically much shorter, virtual desktop performance was similar to a physical PC. However, for remote users with latencies exceeding 100ms, remote desktop performance was consistently slower than a local PC.
Obviously, other remote protocols are available. VMware View 4, for instance, includes Teradici’s display protocol PC-over-IP (PCoIP). PCoIP offers a “seamless desktop experience: rich media and graphics, multimedia and USB” – yet a number of PCoIP users have experienced issues when using PCoIP in a WAN / high latency environment to the point where PCoIP as a protocol for remote access generates a screen that looks like a crayon drawing and is often replaced with RDP. Can PCoIP challenge as a ‘protocol for a complete VDI solution?’ is a question for another day: your pressing question is ‘how do I provide a good working experience for our remote users”?
To Support Users Across the WAN I need Citrix’s ICA Right?
At Citrix’s HDX technologies heart is Citrix’s ICA (Independent Computer Architecture) protocol. ICA is an efficient protocol, long associated as a solution for supporting connections across the WAN. The ICA protocol has a number of features that combine using end device resources and compression to ensure users with high latency network links have a reliable and usable session – typically giving a better experience than VDI solutions built around Microsoft’s Remote Desktop Protocol (RDP).
Building on this experience, Citrix have introduced additional HDX technology components with the latest XenApp and XenDesktop products that can be used to further optimise the rem0te experience for WAN users. The Branch Repeater – available as a physical or a virtual appliance – optimises the ICA protocol through local caching and de-duplication across multiple XenDesktop / XenApp user sessions and local staging of XenApp streamed applications; the Branch Repeaters also sense real-time network and traffic conditions, and dynamically optimize delivery as conditions demand.
However, while Citrix has a reputation for having the best protocol for a Wide Area Network (WAN) environment, the competition has not sat still.
Quest’s vWorkspace 7.1
Quest Software recently released vWorkspace v7.1 0 – a new version of their on-demand virtual services framework. Among a number of new features in the release was EOP Xtream – WAN Acceleration. vWorkspace already included enhancements to RDP that optimised graphics acceleration and reduced the impact of high latency on the user experience. EOP Xtream is a patent-pending technology that further accelerates RDP and Quest’s own Experience Optimisation Pack (EOP) traffic on WANs: it has specifically been designed for users on network links with modest to high round trip latency providing for an improved user experience by providing faster RDP screen responses and improved performance of all EOP features.
An advantage of vWorkspace over other virtual service solutions, such as VMWare’s View, is that it is a framework that works across your desktop delivery services – be they blade PCs, Terminal Services or Hosted Desktops: so it would be possible to manage all sessions connecting via RDP – and optimise the enhance that RDP connection without introducing a different protocol.
While Citrix and Quest both have features designed to deliver VDI services across the WAN both rely on you deploying their framework – be that XenDesktop or vWorkspace. Is it possible to introduce a service that enhances RDP’s WAN performance without changing the framework for managing the desktop services? Yes it is.
Ericom Blaze, for example, is a display protocol that exceeds the performance of the latest versions of RDP – including RDP 7 – across a latent/low bandwidth connection, or for complex graphics or animation displays. Ericom state that Blaze allows you to reduce RDP bandwidth consumption “by up to 25 times – across WANs and congested LANs” by identifying and compressing graphical elements such as bitmaps and key graphical elements, such as the Taskbar and Start Menu and performing packet shaping to further optimize network utilization, and to speed up transmission. This reduces the bandwidth requirement while maintaining the user experience.
Hardware based solutions
Of course, while Ericom Blaze doesn’t require a change in your framework it is another software based solution – you will need to install (and manage) software on the centralised devices and the remote end points. Is it possible to avoid this?
There are a number of vendors that provide WAN optimisation hardware. Expand Networks for example, are the market leader in providing WAN Optimization solutions for Server Based Computing (SBC) environments, they offer a range of physical and virtual devices that can be introduced into your network. Expand’s transparent Real-time All-IP engine can accelerating RDP traffic throughput on average by 300% and can enable 50% more user sessions on the same bandwidth without any changes to the client or server components.
While hardware solutions reduce the need to manage software across servers and devices you will need to install an optimisation device at each branch location as well as the central site. This in turn can increase the costs of your VDI project. To enable more cost effective implementations, Expand Networks do have a Mobile Accelerator Client (MACC) which has different modes of operation to support a wide variety of user working patterns be they home worker, the occasional home worker, road warriors or an entire small office. However, the MACC is a Windows based software client – and it is not uncommon to be considering non Windows devices to be used remote/home working.
Don’t let Poor Network Performance Sabotage your VDI Project
A major driver for delivering thin client (Server Based Computing) was to support remote/branch offices from a centralised location. You got your cost savings from reduced support, reduced hardware costs and importantly reduced WAN costs.
The current raft of VDI solutions focus on delivering ‘PC-like performance’ and this is an important consideration. But, will your solution work as well on the WAN as it does on the LAN?
If you are embarking on a VDI project it is likely you’ll have to deliver services to remote users and as such you may find, like EMC did, that your protocol of choice’s performance across the WAN leads to poor user feedback. You may not be considering a VDI solution, especially one based around RDP, because of RDP’s poor reputation for performance across a WAN
While Citrix’s ICA Protocol cause many to believe that a WAN requirement precludes other solutions, this is not the case. Quest’s latest vWorkspace release’s introduction of EOP Xtream – WAN Acceleration, or Ericom Blaze demonstrates that other VDI management vendors offer solutions to maintaining a good experience for remote users. You can also consider WAN optimisation hardware – but bear in mind these solutions can be expensive as you devices at each location and remote display protocols such as RDP and ICA have different requirements for WAN optimisation – not all devices will necessarily be able to enhance performance and may not be suitable for some user instances – such as roaming users.
Ultimately, don’t allow poor network performance and user experience sabotage your desktop virtualization project. Plan – as EMC did – to validate not only the technical operation of the virtualised desktop, but the user’s experience of using that environment.