Guest post by “San Francisco–Red Hat is the strongest Linux company in the world when it comes to servers, but it has almost no presence on the desktop. That will be changing in 2012 with the reintroduction of a Simple Protocol for Independent Computing Environments-based virtual desktop infrastructure.”
Just what we need, centralized computing. We already have centralized email through a few major providers, the mainstream media, and just a handful of main computer architecture designs.
This is what the establishment wants…to take away your PC and replace it with a terminal, so that they can control your data, your hard drive, your information.
well, when you’re at work it’ s not your computer, not your hard drive and not your information.
His tinfoil hat was personally procured though.
The post is not about a traditional desktop like the summary might have you believe. It is about a thin client setup for business use. Very poor selection for a summary.
Edited 2011-04-12 00:33 UTC
The phrase “Simple Protocol for Independent Computing Environments-based virtual desktop infrastructure” gave it away for me. I actually felt a little thrilled there for a second.
Yeah, especially leaving the “San Francisco” part, making it seem as if RedHat is based there and not RDU.
The phrase “virtual desktop infrastructure” kinda gave it away.
The SPiCE protocol replaces the need for other non-free virtualization solutions. I use it with virt-manager and can cut and paste between VM and host OS.
All we need is better USB 2.x support in KVM and things get really interesting.
I’ve tested SPICE for a bit and also watched Youtube video’s over crappy wifi and even that kinda worked, audio and all. I’m looking forward to using it for all my VM’s.
The idea of SPICE is to use the capabilities of the (this)client if it is available, so if you have 3d-things, the rendering of 3D can actually happen on the client.
That all seems very promising.
To bad development doesn’t go a bit faster with these tools SPICE, libvirt, virtual machine manager. They all could use some polish (?). I think kvm gets the most attention.
Does anyone know if it can also handle individual applications instead of full a desktop ?
(?) Not sure if that is the proper english spelling, but I think you know what I mean.
Edited 2011-04-12 11:10 UTC
They are pretty rough, and they aren’t at a point where they can convince me to switch from VirtualBox with the RDP plugin.
I’m interested in seeing if SPICE could be combined with Wayland to provide a remote access solution similar to NoMachine or RDP.
Is the difference between a thin client device and a full desktop that big (hardware and cost wise) to make it worth creating such kind of hardware?
I mean, a custom thin client device can’t be really cheaper to manufacture than some enterprise desktop based on commodity hardware you purchase from, say dell..
The cost of the hardware is not the main concern. Easy system maintenance, security, data backup, control, easy deployment and relocation are. This is what it brings to corporate environments. And you will use commodity hardware on the client side anyway like you do with Citrix and Windows RDP, and both of them work very well on internal networks and reasonably well over Internet.
That’s not necessarily the point.
For example, let’s say you (major corporation) have a key custom client-server app that you need to distribute to a few hundred suppliers. The suppliers are providing the desktops, but you’re providing the server and data warehouse accessible via an encrypted channel over a wide area network. You have at least three options.
(1) Rewrite the client as a web app. This takes probably a year, includes considerable risk, may affect performance, and may not be feature-complete.
(2) You can send the client to each supplier, work through each of their app approval processes, and provide remote technical support – for every version you release.
(3) You can set up a virtual terminal server such as SPICE or Citrix and run the client on your own servers. To your supplier-based users, it launches like a web app but runs like a native app. You keep full control of the client code, and distribute new versions with a centralized server install. Tech support drops to usability issues, as you know it’s installed correctly on a capable machine. And you needn’t change a single like of code in the client or server to do it.
Really, for rather common scenarios such as this, it’s a no-brainer.
Consider that many people are buying devices like iPads, netbooks and lowend laptops at home. I think desktop systems will get phased out for general use. If a company doesn’t want an employee to have a portal computer, it might be cheaper to buy a thin client rather than a full desktop system in the future because the demand is dropping for them.
Most people are happy with a tinker toy computer and don’t want the power you can get with a desktop or workstation. That could change with the right killer app, but I doubt it will. I’ll keep paying the premium but businesses won’t.
This isnt aimed at the home user. This is aimed at businesses where if I am your boss, I dictate what you need for work. Thin clients are also likely to longer than desktops do.
I know it’s not aimed at the home user. I was making an argument that it might get cheaper for thin clients versus desktops because no one buys desktops anymore!
No it isn’t and existing XP machines can be turned into virtual clients.
Spice is tech that should have came out 10 years ago. Citrix and RDP are already well established. I’ve seen Citrix used over the internet on a junky laptop for some heavy applications and I was really impressed with how smooth it was.
If Red Hat is serious about competing with MS then they need to increase funding to Samba and LibreOffice.
Nope. Our current diskless client boxes are just over $200 CDN. This includes:
* slim desktop case
* M2+ motherboard (onboard hd audio, nVidia 6100 graphics, gigabit NIC)
* AMD Sempron CPU @ 2.0 GHz
* 1 GB RAM
These run full 3D accelerated graphics. No local harddrive, no local optical media, no local floppy drive. Just connect a monitor, keyboard, mouse, and ethernet cable. They are disposable appliances (which is how we treat them).
The least expensive thin-client system we could find (albeit a couple of years ago, but the above system was only a little over $300 CDN at the time) was almost $800 CDN, used a wimpy VIA CPU, had hardly any RAM worth mentioning, and couldn’t handle 30fps video, let alone 3D accelerated graphics.
They could be … if the thin-client companies wanted them to be. But they don’t, so they aren’t. It’s hard to find a thin-client terminal device that’s under $300 CDN. Considering how useless a thin-client device is without a fat server to connect to … it’s even more amazing how much they gouge the end-user.
Especially since most thin-client/remote desktop solutions require a tonne of client access licenses to connect, on top of the client licenses for the software to be run, along with the licenses for the server OS, etc, etc, etc, etc.
A diskless, network-boot setup, along with NX (or Spice, or even an RDP server) with standard, commodity hardware is a *much* nicer setup. You still get all the administrative benefits of centralised installs, setups, configuration, and management … *and* you get all the power of a local “fat” client.
acobar is correct. This is about ease of system administration. Also think security. You know exactly where all your data is and its much harder to steal from a central server room than to steal a random machine, especially a laptop. There were a very few companies actually doing this in 2005 when I was at VMware World. VMware hadn’t even thought that far ahead yet and were still concentrating on the server at the time. IT also makes updating and installing software easy. You can use the same base for every desktop and just proliferate linked clones.
According to my memory, and Wikipedia apparently agrees, Red Hat is not “San Francisco’s” but “Raleigh’s”. Believe it or not, there are tech companies in the US that don’t hail from California.
If there is going to be a focus on the desktop then I think the most valuable contribution Red Hat can make is hiring 50 programmers to work full time on OpenOffice.org/LibreOffice and making integration into GNOME/GTK+3 the top priority. The world runs on applications not operating systems, when Vista failed to get a hold in the business world, was it due to the operating system itself or the application incompatibilities running with Windows Vista? in the long run what there needs to be is a vibrant third party software for the desktop ecosystem for Linux to make it from the technical workstation to the average persons desktop.
Btw, side issue is the wayland server which hopefully will mean an improved experience for the end user once it replaces Xorg on most desktops. The raw ingredients for a great desktop operating system already exist but the problem is the lack of will power to bring that raw material together into a coherent product.
RH went back and forth several times over whether to offer and/or promote Desktop Linux in the late 90s and early 00s.
In 2003 they finally spun off Fedora Project and retreated to the server space.
Now it’s far too late for them to have any impact on the world of Desktop Linux beyond their narrow customer base.