Zero client vs thin client

Zero clients and thin clients are two ways companies deal with the problem of sprawling computer infrastructure its associated costs. Instead of putting an expensive computer on every desk, you put something small and cheap on every desk and share that expensive computer. Let’s look at zero client vs thin client, and the advantages and disadvantages of each.

The VDI model

One place where zero client vs thin client don’t differ at all is reliance on the virtual desktop. Virtual desktop infrastructure, or VDIs, are virtual machines running on servers in your datacenter. The virtual machines run traditional desktop operating systems like Windows 10. All the heavy lifting happens in the datacenter. But the screen displays on your desk.

This is a throwback to early computing, when a mainframe sat in the datacenter and you connected to it with a terminal. The zero client or thin client is the modern replacement for old-fashioned terminals.

First up: Thin clients

zero client vs thin client
This thin client is a small device, not much larger than a mouse, with a keybaord, mouse, and monitor connected displaying an operating system running on another computer.

Thin clients have been around since the 1990s. A thin client is a really cheap and wimpy computer, just powerful enough to connect up to the server and display what the server tells it to display. So a thin client contains a cheap and cheerful sub-GHz CPU (chips based on Cyrix technology were once common) and a tiny bit of RAM and flash storage. In some cases, a thin client has less computing power than a Raspberry Pi.

The thin client boots up a minimal operating system, usually some form of embedded Windows or Linux, which loads up a remote desktop client. The remote desktop client communicates back to the server via Microsoft RDP. The result looks like Windows running locally, but without the big, loud, and expensive CPU.

Thinner than thin: Zero clients

zero client vs thin client
This is a motherboard from a zero client. Note there’s no Intel or anything Intel-like in sight. By TheDukeOfNY [CC BY-SA 3.0 (], from Wikimedia Commons
Thin clients are small, cheap, and dumb. Zero clients take the model a step further, providing something even smaller, cheaper, and yes, dumber. A thin client can work like a general purpose computer, you just wouldn’t want to.

A zero client isn’t a general purpose computing device. It implements RDP in hardware, rather than software, letting you dispense with the flash storage and most of the RAM. The result is simpler, cheaper, more power efficient, and theoretically more reliable, even if it isn’t much smaller.

When you look at the zero client motherboard to the right, you see a couple of memory chips, a network chip, another support chip, some capacitors, and not much else.

How zero clients and thin clients save you money

Zero clients and thin clients reduce downtime. When you virtualize the desktop, nothing stops you from picking it up and moving it to another server for maintenance. The VDI model can improve security too. The way some people implement it, the VDI goes away when you log off, and gets rebuilt on the fly the next time you log on. If the computer gets infected, it won’t stay infected any longer than the user stays logged in.

The other advantage is centralization. When I did desktop support, I couldn’t manage more than about 200 desktops or laptops at a time. We have better technology now so I could scale higher today, but I doubt I could scale beyond three figures. When I worked at a Fortune 20 company using VDIs, they had one person administering thousands of VDIs. They stretched him further than I would have recommended, but centralization increased his productivity an order of magnitude beyond what I was doing in the 1990s. That saves you a lot of expensive labor.

VDIs also save you a lot of hardware cost. A server powerful enough to replace 100 desktop computers costs less than 100 desktops. I also don’t have to dedicate as much storage and memory to a VDI as I would have to dedicate to each desktop. When storage comes in 128GB increments, you can’t dedicate less than that to a single machine. But with VDIs you can.

There’s an unintented side effect too. Getting rid of desktop PCs reduces shadow IT, because if there aren’t carts of desktop computers laying around, you can’t steal them and turn them into rogue servers.

The problem with VDIs

I’ve never met anyone who likes their VDIs. They tend to be slow and unreliable and hard to administer. I think it’s mostly because people overprovision too much, trying to squeeze too many VDIs onto a server by skimping too far on storage and RAM. But if it were that simple, someone probably would have solved it.

There’s a big opportunity in this space for someone. Companies recognize the need for VDIs. They buy them and use them even though they hate them. If someone can get VDIs right, they stand to make a lot of money.

Zero client vs thin client: Two paths back to the future

In computing, there are no truly new ideas it seems. More often than not, we cycle through old ideas, just applying a new spin to make an obsolete idea solve tomorrow’s problem. The VDI model is the throwback to the minicomputer or mainframe in the datacenter serving terminals on the desktop. Zero clients and thin clients are just two different modern replacements for that old terminal.

3 thoughts on “Zero client vs thin client

  • May 2, 2018 at 10:18 am

    People overprovision because that’s where the cost savings come from. If you install a powerful enough server to actually provide good service to everybody and enough network infrastructure to handle all the RDP traffic well, you don’t save any money with thin or zero clients, at least not on hardware. It might still be worth doing because of the easier management, especially in environments where data privacy is important.

    • May 2, 2018 at 11:18 pm

      I agree 100%. I also agree it would be worth doing even if you didn’t save on hardware for exactly those reasons. I’ve seen how stretched thin desktop support resources are. And data is a company’s most valuable resource. I can think of one company I’ve worked with whose business is all B2B so they’ve never sold anything to you or me directly, but if their data were stolen, they could lose competitive advantage.

  • May 3, 2018 at 6:46 am

    Many organizations will have some users who are edge cases that make them poor candidates for using a thin client. Some possible reasons:

    Use hardware that is not present (such as high end GPUs) in the standard configuration
    Need to run software that requires administrative privileges
    Have a job where visual quality is crucial (photo or video editing) – RDP compression compromises that
    Have a job that routinely involves installing software (product evaluation, competitive analysis)
    Need a type of computer that is not otherwise used in the business (Mac or Linux in a Windows shop)
    Use multiple displays (multiple display support in thin clients is improving but may not be sufficient for all)
    Make intense use of video (video editing, game development)
    Interface with hardware with no network capabilities (scanners, lab equipment, badge printers, etc)
    Develop hardware (computer is interfaced with things under development)

    In some cases, it may make sense to give those users a thin client that is used for routine job functions (reading and sending company messages, logging time worked, etc) as well as a separate computer that is only used for running things that fall outside the capabilities of the thin client. It’s especially important to get IT staff to use the thin clients for some of their work, even though other parts of their job requires doing things that are outside its capabilities, so that they will be familiar with its capabilities, performance, and limitations.


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this:
WordPress Appliance - Powered by TurnKey Linux