If you have dismissed virtual desktop infrastructure (VDI)
because you believe server-side graphics sharing is a myth—like unicorns
or measurable Windows Phone market share—Nvidia Corporation wants to
introduce you to Grid 2.0.
Announced at VMworld 2015 in San Francisco, Grid 2.0 is the
next generation of Nvidia's virtualised graphics technology, and you can
get your hands on it today. And, no, this isn't about its streaming game technology of the same name.
If you're new to virtualised desktops, think of a
virtualised desktop as an umbrella term that refers to a multitude of
ways IT can run a workload, an app, or even an entire desktop on a
server—but display it on a client device as though it were actually
running there. The hiccup has always been masking a graphics experience
that's been bottlenecked by the network; pumping graphics to a multitude
of users across a 100Mbps or even 1Gbps shared local network pipe is
much slower than zapping pixels from a graphics card to a display in the
same system—especially when it's a few dozen users demanding video from
the same struggling server. VDI employs a variety of technologies to
streamline that graphics experience such that it can flit down a network
pipe and users are none the wiser. Grid 2.0 is Nvidia's latest solution
to this problem and it's impressive.
New Levels of Graphics-Intensive Applications
With this version, Nvidia claims Grid 2.0 will deliver new
levels of graphics-intensive applications to any device using
virtualisation—not just to VMware's vSphere but to other standards, too,
from various companies including Microsoft and Citrix. To do this, Grid
2.0 is supposed to double performance from version 1.0 and increase
user density to 128 users per Grid-enabled server. It's also been
expanded to support Linux as a target operating system (OS) and can be
run from blade server arrays in addition to larger rack boxes.
Using Grid 2.0, customers can virtualise applications and
even entire desktop environments, and deliver them to different kinds of
connected devices—especially thin clients and small-form-factor mobile
devices. That means not only can an $85 thin client run a full Windows
desktop with support for Aero and 3D graphics, theoretically an iPad
could, too. At VMworld, Nvidia is showing off Grid's new muscle with a
"Tower of Power."
Despite its cringeworthy name, the Tower of Power's display
is impressive. It's 487cm tall, with all four sides displaying 336
micro-tiles grouped into 56 desk-sized displays—and not just displaying
but occasionally showing advanced special effects cascading across the
entire tile scape, including tornados, ripples, and I don't know what
else. This was a truly impressive display considering the whole thing
was being virtualised server-side.
Native Management
If you employ Grid as a support layer for a third-party,
server-side VDI system (such as Microsoft RemoteFX), then it will
seamlessly support those management tools, according to Nvidia. But Grid
also comes with its own management toolkit that lets IT deploy and
manage virtual desktops. Using those tools, they can establish user
profiles to appropriately allocate server-side graphics resources and
deliver the right level of performance to each user.
Nvidia announced partner support for Grid 2.0 from the likes
of Cisco, Dell, Hewlett-Packard, and Lenovo, claiming those worthies
will be offering Grid 2.0 on over 125 server models. You'll see an
official release around September 15 but Nvidia also has a demo environment available using VMware vSphere 6.
Post a Comment