Dumb terminals were common in the age of mainframes, but they faded upon the rapid proliferation of desktop computers.

Nowadays most companies own modest datacenters that take advantage of virtualization to support hundreds of machines, and cloud services are more popular than ever.

Are we advancing towards the rebirth of dumb terminals in companies? What are the currently available solutions on this matteR?

  • 4
    "The internet has progressed from smart people in front of dumb terminals to dumb people in front of smart terminals" - I forget who said it, but it rings true. – Paul Tomblin May 29 '09 at 12:06
  • I would add "more", therefore "the internet has progressed MORE from smart people in front of dumb terminals than from dumb people in front of smart terminals". Don't know who said it either. – Rook May 29 '09 at 13:12

I did a product evaluation for a call centre. I looked at deploying 50+ terminals on an exposed ground floor office. Using VMware Virtual Desktop Infrastructure and the starter packs : 3 back-end servers. Costs came in around 30% cheaper than buying 50 new desktops.

As regards deployment - deploying 50 desktops on VMWare VDI would have taken about 2 days. One day to prepare the image we wanted, and another day to convert it to a VMWare image and set up the deployment transformation scripts.

Final analysis was that VDI would mean a cost of less than 50% of the cost of buying new PC's and getting them set up in the traditional image/sysprep method. Added advantages of centralised management, easy patching, simple to extend the capabilities to homeworkers using VPN, and no data on the desktop (in case of ram raid and IT equipment theft)

Added IT deployment bonus was that we could set it up in under a week and be ready to drop Wyse terminals on desktops the day people moved in, rather than having to buy 50 PC's and set them up leaving them in an empty exposed office.

We also found Wyse terminals at £200 were a lot more cost effective than desktops at £400 - also, its only the screen they display, the back-end grunt can be upgraded centrally by beefing up the servers, so less IT equipment churn, an a lot lot less PC envy. ;)

So all in all - I think that centralised computing with dumb terminals on the desktop is coming back - but this is not the first time it has returned. We will continue to centralise, decentralise and so and and so forth until we have the computers embedded in our brains with a hive mesh wireless network to control us all? Was JRR Tolkein actually talking about token networks ????


Dumb terminals have long returned in the incarnation of the thin-client. Bit different, but much the same.

Current solutions include VDI (Virtual Desktop Infrastructure) from a host of different vendors (ranging from VMware to Red Hat and way beyond that), Citrix, Sun Ray... I could keep going for quite some time.

If you ask me, it's a circle. Sure, the cloud will have it's day. Just until Intel figures out a way to get people to move back to individual workstations (=profit).

We've moved half-way around, from mainframe to pc to cloud now. It'll not happen tomorrow, but sooner or later the wheel will continue spinning and initiate the move back to more powerful workstations. Maybe it'll be a resource hungry OS, maybe it'll be cheap and green power, maybe it'll be the realization virtualization is not the egg of Columbus either. Who knows.


The problem is that most of the new "dumb" terminals like Wyse thin client terminals are actually quite expensive. It can be hard to persuade management to buy machines with reduced functionality compared to a desktop PC of almost equivalent price.

  • One way to 'sell' a thin-client/dumb-terminal solution to management is to take into account the TCO of the machines. Thin-clients can usually be configured so that the host os/configuration cannot be changed in any way by users. This Is a great way minimize support costs. (Sell the support savings with the initial purchase price..) – Peter Bernier Jun 23 '09 at 14:24

I've also used thin-clients in PC Hazardous environments. I once ran the network of a industrial rubber manufacturer. The factory was extremely dirty, and absolutely horrendous on PC's. So, when I needed to replace the old 5250 twinax terminals, I went with thin clients. No drives = no moving parts = longer life. Plus it gave me something to do with all the CRTs I was taking out of the office cubes but could not get rid of.

  • I agree whole-heartedly. In any sort of a plant environment, thin-clients are the way to go. Compared to the cost of 'industrial' PCs, they're also cheaper. For example, we use both for certain applications. Our thin-clients cost about $600 a piece, whereas the industrial PCs are upwards of $1600 depending on configuration. (Yes they're used in completely different applications, but if you don't need the hardware, a thin-client makes more financial sense...) – Peter Bernier Jun 23 '09 at 14:26

Back in 1980, a dumb terminal might have had a Z80 processor or something similar running the terminal's embedded software. This meant that the hardware cost a few thousand dollars instead of the cost of a PDP-11, Vax or mainframe. However, this technology had mostly died out by the end of the 1990s.

Nowadays, the cost of a PC CPU is such that there isn't much advantage to making dedicated 'dumb' terminal hardware in anything but highly specialised instances. In some cases (such as POS machines) the handset is actually connected via a serial link and works like a serial terminal.

Some companies such as WYSE still make thin client hardware that have x-server, citrix or RDP software. There is still a market for this type of machine, although they are not much less expensive than a cheap PC. Sometimes this type of hardware has serial ports and a terminal emulator so it can pretend to be an ordinary serial terminal.

Arguably, the browser acts as a terminal interface to a remote application. In many ways it acts like an old-style 3270/5250 type block mode terminal, submitting a completed 'form' request to an application.

The most common solution these days is to run terminal emulation software or remote desktop connections (depending on the type of host) on a general purpose PC. There are few situations where this is not the optimal solution - in most instances the user of this software also uses general PC software such as word processors or email.


The Control -vs- Capability argument has been going on for years.

The biggest factor seems to be the network. Many of the "Terminal style" systems available move everything, including drawing to the screen to the server, requiring the entire screen be sent across the network.

Another factor is energy costs. Many computers capable of running as "terminals" use FAR less power than a typical desktop PC. (10's of watts instead of 100's of watts)

As networks become more powerful, and energy costs go up, consolidating processing becomes once again a viable alternative.


Yes, we are. Developing web applications is simpler than desktop applications (for many tasks). But the terminals have become less dumb.

These days thin client means RDP / Citrix / X / Web browser.

  • Interesting. I find it much harder to do the same tasks in a web app than a desktop app. The web development "libraries" are very weak in comparison to what's available on the desktop. – Brian Knoblauch May 29 '09 at 11:34

Today's dumb terminal equivalent is the BROWSER! If you were to look at a typical 3270 terminal app and a current html forms based, you'd' notice that the operational semantics are the same. The key has been replaced by a push button.

Yup! thin-clients are expensive, Citrix, etc often an overkill. I find the new low power devices with no fans, e.g. NorhTec MicroClient a cheap and effective option in industrial settings. I just put gaff tape over the unused connectors and the CF slot.

Not the answer you're looking for? Browse other questions tagged or ask your own question.