Cloudy desktops are as mature as cloudy servers … from 2008!
It's not a DaaS-aster, but not better or cheaper than VDI, yet
Desktop-as-a-service in 2016 is about as mature as infrastructure-as-a-service was in 2008, so waiting until it matures is more sensible than diving in now.
So says Garter for Technical Professionals' analyst Mark Lockwood, who The Register's virtualisation desk beheld yesterday at the firm's Infrastructure Operations & Data Centre Summit in Sydney.
Lockwood said DaaS currently lags desktop virtualisation (VDI) in many ways, especially on cost. Best-practice VDI costs about US$300/seat/year. DaaS costs more. VDI doesn't have latency problems. DaaS does and those problems only get worse if your desktops have to come in over the WAN to reach data inside the firewall, pipe that data into the cloudy desktop and then send it to users over the WAN again.
Current DaaS offerings are also a little unrealistic – the base configuration of a single CPU with 2GB of RAM is not useful for most applications. More realistically-specced machines cost about US$50/month. Suppliers are also thin on the ground. Today only Amazon Web Services and VMware provide a single-throat-to-choke experience. Other providers can split bills so you pay for VDI licences and for the cloudy desktops.
On the upside, Lockwood said VDI is a notoriously finicky application that requires a big up-front investment and even then often needs tweaks. In typical cloudy style, DaaS requires no up-front spend. DaaS' elasticity is also helpful for businesses that employ seasonal workers. With on-premises VDI you need to buy for peak user numbers and then wear the cost of idle infrastructure. DaaS is also nicely secure: if a mobile worker uses a cloudy desktop on a properly-secured device, sensitive data should be out of reach to whoever it is finds a laptop or iPad in an airline seat..
Like all cloud services, DaaS will of course get better and cheaper. But Lockwood said there's no point in waiting for DaaS to improve if VDI is something you need now. That's because DaaS is still rather immature, but just as IaaS has come a long way in a short time Lockwood expects it won't be long before the list of reasons not to consider DaaS becomes rather shorter. ®