Last month Barry Schiffer and I did a presentation on E2Evc, the subject? “Enterprise clients. when to use what and why?”
The last couple of meetings I had with customers this was a very hot topic, people are rethinking their client strategy with BYOD and CYO as alternatives to their traditional desktop strategy. I noticed that there still is a demand for a client strategy when using SBC/VDI but there are a lot of choices to be made when looking at enterprise clients:
| Thin Clients | Fat Clients | ||||
| Zero Clients | Linux Clients | ||||
| Linux Clients | Windows Clients | ||||
| Windows Clients | Domain Joined | ||||
| Domain Joined | Non-Domain joined | ||||
| Non-Domain Joined | Antivirus | ||||
| Antivirus |
And this is just the start, Andrew Morgan did a presentation too and came up with the following discussion points:
Linux:
- Hardware drivers, smart cards, printer drivers and scanning have their downfalls (although driver support is getting better specific driver support for linux isn’t always widely available)
- Windows Admins are notoriously bad at administering linux and will roll back to a known platform if they have a problem (Golden rule: If it ain’t broke..)
- Linux is lower in priority for Citrix feature releases.
- Company policy: Windows Only
Windows:
- Familiar and comfortable for windows administrators.
- Can be managed by current patching and AV solutions.
- Few to no driver issues.
- Local applications will more than likely work (in combination with RES VDX or Citrix’s project Rainmaker).
As Barry and me where looking for an approach for this presentation Barry came up with the analogy between thin clients/fat clients and woman.. Guess we all know where that came from.. 😉
We used the following slides:
Where the pro’s and con’s are explained in the text for both a thin and fat client but we did some testing ourselves. We got some test devices and measured power usage as an example to see if a fat client would use that much power in order to deliver the performance (Thanks Lenovo, Wyse and iGEL!).
In comparison (not with our tested models btw) you can see that a device like the Lenovo which is the only fat client we tested is relatively small. The second device from the left is a Lenovo Tiny, it’s the first Ivy Bridge fat client in the Netherlands so thank you IBM/Lenovo for lending us this device.
In the picture above we showed our test devices, one of them is a fat client. Can you guess which one?
We setup XenApp 6.5 and used a manual test (Big Bunny HD anyone?). We wanted to test the devices using an automated test but because it usages a launcher we couldn’t use it on the Linux/Zero clients. The outcome of our test was rather surprising:
As you can see in the picture above the Lenovo client didn’t use a lot of power but it could give us the best performance of all tested devices. From Lenovo we got the following results and as we can see we have similar test results although the testing methods where probably different:
- DevonIT VDI Blaster
- Microsoft ThinPC
- Citrix Desktop Appliance lock
- IGEL Universal Desktop Converter
- RES Software Subscriber
The problem I found in these solutions is that they’re not easy to configure and when configured they’re hard to adjust. Some of them even use USB keys to boot from and the disadvantage of USB keys is that they tend to ‘get lost’ over a period of time. Sometimes this is configured throughout the project and admins won’t or can’t touch it because of lack of knowledge, experience and documentation.
One of the exceptions to these disadvantages is Andrew Morgan’s ThinKiosk (free tool), his goal for thinkiosk:
- Central management.
- Ease of configuration.
- Power Management.
- One size fits all for all desktop models you publish to your users
What else did he want his solution to provide?
Native support for HDX functionality:
- HDX Flash Redirection
- HDX Multimedia Redirection
- Branch Repeater Plugin
- Empower user based configuration changes when needed.
- Support for applications or multiple desktops.
- roll back to the users fat client desktop.
So our conclusion was: There’s a market for both but as thin clients/zero clients normally are under scaled in hardware perspective and the price/quality isn’t always what it should be I see more and more deployments of fat clients that are being reused as thin clients. If you want a feature and future proof, low cost client you should consider a device like the Lenovo Tiny with a tool like ThinKiosk to leverage all functionality but simplify management.
[Updated]
The E2EVC organisation just released the HD video’s from the presentations, so I added ours:
Kees Baggerman
Latest posts by Kees Baggerman (see all)
- Nutanix AHV and Citrix MCS: Adding a persistent disk via Powershell – v2 - November 19, 2019
- Recovering a Protection Domain snapshot to a VM - September 13, 2019
- Checking power settings on VMs using powershell - September 11, 2019
- Updated: VM Reporting Script for Nutanix with Powershell - July 3, 2019
- Updated (again!): VM Reporting Script for Nutanix AHV/vSphere with Powershell - June 17, 2019
Re-using current Fat Clients has several advantages as you already stated. Especially when they are domain joined and you can use WSUS, policies and other products already available to install, configure and maintain them. Till know every customer still needs some real Fat Clients, so you need to have the infrastructure available for those and then one management solutions is easier then to have to (one to maintain the thin clients and the other for the Fat Clients).
With some simple scripting and GPO’s you can configure very easy a Fat Client as Thin Client (although ThinKiosk is even better) as described in this article.
http://sbc.vanbragt.net/mambo/index.php?option=com_content&task=view&id=1648&Itemid=191
Hi Kees,
Good article and even better question, is there a future for thin clients?
I guess it all depends on the definition of a thin client. How thin is thin?
Thin clients used to have the following advantages;
– low purchase cost
– longer durability (because there where no spinning parts).
– easy maintenance
– acceptable set of features
– small form factor (less space required)
– easy to use
I mean, that’s how we sold them to our customers. Right?
But as you explained this ideal thinclient device turned out to be a unicorn, it never met the requirements of the ‘pyramid’ (see below).
The big advantage of a ‘fat’ client (how can we still call the Lenovo Tiny a fat client, have you seen that slick MTF!?!) is that it has components that will (probably) last longer then the features they need to deliver.
What I mean is; with an Intel i5 you probably can run all HDX features (even those who are coming) in a few years. Where a thin client is outdated the moment it comes out (sure you can upgrade a thin client, but did you mention that in your business plan)!
In the end the quality of the device is found in a perfect balance in the pyramid:
– The user wants great user experience (HDX features?);
– The admin want easy maintenance and control;
– The management want low costs and maintenance.
I like the concept of the Lenovo ThinkCentre M92P Tiny as a replacement of a thin client but it has big disadvantage;
– the initial cost to purchase the device (quit contradictive to Tiny);
– the license cost for the OS (Windows)
– the license cost to manage the device (SCCM license.
Cheers!
Ingmar Verheij
@ingmar:
– the initial cost to purchase the device (quit contradictive to Tiny)
That’s why I said ‘a device like’, there are other configurations you can choose that can replace the Tiny. I just used it as example because we had a test unit at hands.
– the license cost for the OS (Windows)
When OEM-ed we don’t have any problems in costs if the customer has SA we could deploy ThinPC under the SA agreement.
– the license cost to manage the device (SCCM license)
Why should we need a SCCM license to manage the device? We can use GPO’s or RES Automation Manager and lower the costs to almost nothing?
@Kees:
– Regarding the costs of the device; If other configurations exists with the same benefits, that’s not an issue. But I haven’t seen such a small SFF with those specs. But the initial price is always higher than the price we’ll eventually pay (let’s hope so).
– If the customer has SA I agree, but not all customers have SA. Companies without SA would rather buy a thin client with a proprietary OS to keep the cost down.
– SCCM was just an example, the use for RES Automation Manager isn’t free of charge either. There still are costs involved.
Most management can probably be done with GPO’s (but who want to, it’s crap) and prevented with ThinKiosk (which is great!).
@Ingmar: Why would we need a small SFF? We can do with a normal desktop too? Could it be that size is just a ‘nice-to-have’? As for the licensing part you’re ignoring my comment on the OEM-ed workstations so there’s always an OS to start with?
Great article and nice discussion!
“Linux is lower in priority for Citrix feature releases” this is true, but Citrix is also closing the gap more and more between the Windows Receiver and the Linux Receiver.
A lot of HDX features (like 3D codecs, flash redirection, etc) is available on the Linux Receiver and also HDX SoC is coming, but indeed some features of the Windows Receiver are not available on Linux, and maybe they never will be because of the difference in operating system.
I agree on the discussion points about Linux, but if this is not an issue, i think for a lot of use cases the Linux thinclient (like the HP Smart client) is more then enough to provide task workers functionality for there day-to-day work.
If we want more functionality (like Printer redirection, etc) we need a Windows based client device, because Windows is not a OS that is made for hardware that is used by thin clients now a days (this is coming with W8 RT) but if we look at here and now, then i agree there is no future for Windows based thinclients.
So at the end i think we cannot speak of the term “thinclient” when there is Windows running on it, only because Windows is running on some kind of small form factor hardware doesn’t make it a thinclient IMHO.
Thinclients IMHO are arm-based small footprint devices that can be remotely managed through a network service.
So we have two types of static SBC client devices (with static i mean devices on the desk and not BYOD and MAC’s etc):
1 : Arm based thinclients for 80% of the task workers
2 : Windows based fat clients for the other use cases (refurbished PC’s or in tiny form factor and managed with thinkiosk 🙂
Kind Regards,
Bram
Looking at this as a proponent of thin clients, I’d say there are environments where thin is the ONLY way to go. Any other environment it is a toss up depending on your network architecture and end user needs.
Right off the bat, any industrial application begs for thin clients. PCs and most fat clients just aren’t built to deal with constant vibration, airborne particulates, or high heat. And if you are unfortunate enough to have a PC in that environment, the only one happy about it will be the IT guy because it will provide him with job security since everything will keep breaking on a weekly basis.
Also, I am not a fan of spending 2 days rolling through a 100 client network playing “plug and chug” with individual machine updates. I would much rather install one update to my one paid license on a Thin Server, and have it distribute that updated version to every client running a session without having to sit through 8 minute long reboot cycles.
From there, it is easy peasy to set individual application access levels directly from the server.
Now don’t get the wrong idea here…ThinManager can turn a PC/fat client into a thin client with the standard WinTMC feature which gives you great flexibility in available architecture, but you are still spending more to have a PC do the job of a $129 thin client.
Bottom line…get the OS off the desktop and save yourself time and aggravation.
Guys,
I love the concept of a “Windows Thinclient” like device. The only downsside is that this solution needs more management. Security patches etc. It’s not difficult but still it needs more management. Most companies have a solution in place for this and then it is not a big issue. At my current customer we use this technology to reuse old PC’s and still have good HDX performace on an older dual core Intel processor.
Hi Kees,
Great article and thanks for the mention.
From my point of view, this clients and their use cases dictate the model to use. If you need one a one size fits all your clients and accept the cost impact of this, then the Lenovo tiny or a similar model would be fine.
But if you want to be cost effective I think a mix of these solutions will make the most robust solution. My decision tree would look something like this:
1:we have enough hardware already:
ThinKiosk for all
2: We need more hardware:
Is the user a task worker that will never need redirection? > zero client
Will the user need redirection? > pc with ThinKiosk.
Is the user a developer / graphics person / etc? > local pc for now, re-assess when xendesktop remote pc is released.
I do understand the discussion, but every customer and user group have different requirements. Thin clients is just one solution, just like sbc or vdi is just one solution. You can also question why to stick with sbc or vdi when you are putting full performance pc’s on the users desk? You are allready managing the pc’s so push out tyhe software too!
On the other side you missed the top thin clients currently out with an amd t56n low power cpu with integrated radeon hd graphics available in wyse and hp thin clients like the hp t610.
A 30watt differene for 5000 workplaces is still 150 kwatt difference * 8 hrs * 5 days a week * 47 working weeks a year…
[…] article is not about the FAT Client ThinKiosk vs Thin Client discussion. Please check Citrix: Is there a future for thin clients? for […]