The Peculiarly Quiet Decline and Fall of the KVM | Hacker News Reader

The Peculiarly Quiet Decline and Fall of the KVM

loper-os.org

132 pointsblfr138 comments

KVM switches used to be a huge and super important (and expensive) thing in rack server farms back when every rack or every few racks had a pull-out tray with a keyboard and a monitor. Technology moved on. First the tray and monitor disappeared with KVM-over-IP, which saved many trips to the server room. Then the KVM switches themselves disappeared, as manufacturers integrated out-of-band remote access into the servers, first with add-on boards, then onboard and finally on-chip (this is what Intel AMT does, when it isn't busy running rootkits).

The little two-PC switch box the article refers to (I think) was always a spillover from the server KVM technology, except with a very small market and more difficult to make. Servers don't need sound and are fine with just VGA video. Home computers want USB (which variety?) and audio switched, or high bandwidth signals that are very touchy about interference or suddenly being connected to a different device like HDMI, Displayport or Thunderbolt. And then people want the stuff for cheap.

> signal switches with arbitrary bandwidth remain available: analogue relays

Have you tried that? Cut a HDMI cable in half and splay it out, then try switching that to another display. See what happens. If it were that easy, it'd be around (or you'd just be building it yourself).

As well the use of KVMs by PC enthusiasts has most likely been replaced by virtual machines.

Yes, no doubt that rack-based kvm stuff disappeared. It was awful: hard to rig, hard to use with confidence.

(Confidence? "Wait, wait, am I talking to data2 or web4 or what?)

KVM was a just-barely-competent solution to a hair-on-fire (=== must-solve-now) problem.

The easiest solution is to just buy an HDMI switch(there's loads of them, even ones which work with 4K signal, and they don't disrupt DHCP) and a USB switch. Total solution should fit in about 40-50 dollars.

But, that's more than one button to push! /s

Seriously though, I am a regular user of DVI and VGA KVMs, which have served me well over the years, but now as I migrate to HDMI or DisplayPort connectors and higher resolutions, I'm finding myself having to re-think how I hook everything together.

My windows laptop has only VGA out, so I hook up a second monitor via displaylink (usb -> DVI).

I tried plugging the cable into my KVM switch and it worked okay on the Windows 10 side but there was some hot-plugging kernel panic when switching to Linux (must file a bug report!)

So perhaps the future of KVM switches doesn't need video connectors such as VGA/DVI/DP if USB-C -> HDMI (as smartphones are adopting) becomes universal.

Are there any that don't suck bogwater? We've tried a handful for our conference rooms, and they are all universally flaky, with signal dropping randomly all the time.

I found that HDMI switches lose signal because the output device doesn't provide enough power for the switch to leach off. I've had a few of Aukey switches: https://www.amazon.co.uk/Switcher-Output-Splitter-Control-Bl...

And once you plug in a dedicated 5V power supply they are rock solid.

That one doesn't meat the (IMO ridiculous) requirements of TFA, which is 4k@60Hz

I have a passive switch that works fine. It's pretty much equivalent to unplugging the cable and plugging it in to a different source/output.

http://www.cablechick.com.au/cables/4-port-hdmi-switcher.htm...

(I have also tried a simple powered switch with an IR remote control, but it caused signal degradation & glitches.)

About 25 total actually (I use these both for my 4k setup)

https://www.amazon.com/gp/product/B071CJWSPF/

https://www.amazon.com/Plugable-One-Button-Swapping-Between-...

4k 60fps 4/4/4 color no problem

Anecdotal, I still use a KVM, for a specific use case. My PC has 2 monitors. My 1st monitor is standard usage, but my 2nd monitor is also used by my partner's PC. If she wants to use her computer, she gets the 2nd monitor (and the speakers) but otherwise I got a 2nd monitor to use.

Practically, it saves me from having to use another set of speakers (you would not use 2 sets of speakers at the same time and I usually use headphones) and it also saves me from having to have or use 3 monitors which takes up more space and costs more.

My partner doesn't use her computer a lot. When she does, all the hardware is available for her. I use the 2nd monitor during development, gaming, watching movies, and RDP session.

As per some comments, I don't see how a VM would replace this use-case but I do understand its niche. Still, it saves us space.

You are not alone in this. For years my Partner and I had separate machines on a KVM. It was only when our Daughter moved out and her desk became free in our study that we no longer needed this setup.

You had a daughter with your business partner? I'm now curious about the story.

Partner can be used in lieu of spouse for an unmarried couple.

Even for married couples!

Thanks for the pointer. I wasn't happy with my phrasing. I should have just referenced a dictionary. MW's sense d is what I intended to convey:

> "a person with whom one shares an intimate relationship : one member of a couple"

https://www.merriam-webster.com/dictionary/partner

I use that term because its gender neutral and relationship neutral. I find it irrelevant to a discussion whether we are married or not, or a member of the LGBT community or not. However I don't hide my gender or my partner's gender (ie. I use he and she forms due to practical reasons).

I have always assumed "partner" either means "business partner" or that the person is signaling both (1) "I'm gay" without being overt and (2) "I'm resent society's prevailing idea of marriage." at the same time.

One can use "spouse" or "significant other" to avoid the perception of being passive-aggressive.

I'd encourage you to give others the benefit of the doubt. Without other evidence, that "perception of being passive-agressive" is only that, a perception.

Those terms work as well, but the term "partner" translates 1:1 in my native language so I can use the term in my 2 main languages. YMMV.

Don't your monitors have multiple inputs? Mine do and have for some years.

I think both of my current ones have built in (crappy) sound and both HDMI and Display-Port inputs.

Fair point, but I had this KVM lying around as I bought this back in ~2007 when I was using a Linux server as 24/7 desktop (including even some Windows apps via WINE) as well whereas I put my Windows desktop off. I was using one keyboard/mouse/speakerset. While I'm not using the keyboard/mouse function anymore, we are using the speaker function.

Also, that feature you mentioned, while it works with some RCs (generally "source" or "input" key), doesn't take into account all the inputs which I am not using. So I need to go through a list of 4 or 5 devices including DVB-C cause yeah my monitor has that support even though I've never used that input. That's annoying.

It gets annoying the more often you gotta switch. For employees of government (like the mentioned example of the Pentagon) that's gonna be more often than not. Any downtime is then a waste because even if its merely 20 seconds back and forth it quickly adds up.

It could even make sense to have KVMs in businesses where they want to keep certain data strictly internal. But as long as companies aren't being punished for data leaks that doesn't have priority. Perhaps we'll eventually see a rise again of KVMs.

Yeah, got to agree it's far from perfect - I use it for infrequent direct access to the server under my desk, and it does mean messing around with having extra input devices etc.

> Cut a HDMI cable in half and splay it out, then try switching that to another display. See what happens.

If you took an ancient (but still available to buy) 25-pin 'parallel port' 4-port switch box with a rotary dial:

https://photography-909a.kxcdn.com/product-images/03292/600/...

And wired up the HDMI of two machines to output ports 1 & 3[1]. Then wired up the HDMI to the display to the input port (so effectively using the switch 'backwards'). As you are just simulating unplugging and re-plugging in the HDMI cable from each machine, via the switch, then surely this would work?

There would be a bit of delay after switching as the HDMI/HDCP handshake completes, but unless you are switching every few minutes it should be usable?

If it does work, then you've got 6 pins in the switch left to play with which could be used to switch usb devices (via a downstream hub)

I've not done this, but have thought about it for a while now.

----

[1] Not ports 1 & 2 as you want to ensure total electrical isolation between the devices during switching.

I think the real isssue is with HDCP.

The output device "ensures" the display device is authorized via a key to output.

So, you would need an intermediary device that would accept any incoming signal, while also having a legit handshake with the output device.

So, as you switch "inputs", there is no additional authorization required, because the output device does not detect any changes.

I believe this is also the case. The only organizations that can get exemptions from the HDCP Alliance (DRM Cabal) for devices that effectively circumvent encryption are for high end broadcast gear. Nothing precludes making a 4k KVM. Infact, a mid range AV main unit could probably route 4k KVM.

https://usa.yamaha.com/products/audio_visual/av_receivers_am...

It's absolutely not HDCP since a 4k HDMI splitter is much cheaper than a 4k KVM.

My guess is that with KVMs disappearing from server racks, there isn't enough volume to get the prices we used to get.

Would it have enough bandwidth? High speed connections have some careful engineering (for instance, maximum distance without a twist on a differential pair), that I doubt that ancient box has.

Also, there are sequencing requirements. Look carefully at a USB-A plug, and you'll see that the power and ground traces are longer than the data traces: when plugging the USB cable, the power is connected first, before the data; when unplugging, the order is reversed. HDMI probably has similar requirements, perhaps with more steps. Again, I doubt that ancient box can do that.

Not likely. If the cables are short you might get away with it but not likely.

The problem you will have is that the signals require high bandwidth and the cables are carefully designed to have low crosstalk, high immunity to noise (both via twisted pairs) and a 100 ohm characteristic (dynamic) impedance (how the cables are constructed, insulation thicknesses, how they are twisted).

DB-25 switches work for simple single ended low frequency signals. The physical switch will wreck havoc with the interpair skew, far-end crosstalk, attenuation and differential impedance.

HDMI 1.3 defines two cable categories: Category 1-certified cables, which have been tested at 74.5 MHz (which would include resolutions such as 720p60 and 1080i60), and Category 2-certified cables, which have been tested at 340 MHz (which would include resolutions such as 1080p60 and 2160p30). Category 1 HDMI cables are marketed as "Standard" and Category 2 HDMI cables as "High Speed". This labeling guideline for HDMI cables went into effect on October 17, 2008. Category 1 and 2 cables can either meet the required parameter specifications for interpair skew, far-end crosstalk, attenuation and differential impedance, or they can meet the required nonequalized/equalized eye diagram requirements. [1]

[1] https://en.wikipedia.org/wiki/HDMI#Cables

Thank you for the clarification! :)

I actually bought a 4-port USB/HDMI switch model fairly recently. It's really just a clutter and convenience thing. I have a few computers, including Raspberry Pis, that I just fiddle around with sometimes. While I could do things in different ways, it's just simple to have one monitor, keyboard, and mouse and plug the computers into the switch.

Do you have time to link the specific product if it has been working well for you? I personally wouldn't even mind a referral link though most get downvoted, especially if they lack a disclaimer.

Something that can reliably connect to random/miscellaneous projects would be really useful to me!

This is what I got: https://www.amazon.com/gp/product/B06XRDS1CG/ref=oh_aui_deta...

It seems to work fine although I haven't used it a huge amount.

This does not meet the requirements of a modern workstation by my or the original article's requirements. This is limited to 1080p. We're looking at 4k as a minimum.

There do exist some 4K and even freesync capable switches with USB3, I only know of the Level1techs ones

Single Monitor https://store.level1techs.com/products/kvm-switch-single-mon...

Dual Monitor https://store.level1techs.com/products/kvm-switch-dual-monit...

At work, we have at least the first 30 racks with good 'ol VGA KVM's (no IP) and tray monitor/keyboard/touchpad, despite at least half of the servers having IPMI. Ain't going anywhere anytime soon.

And no, no one needs 4k where a KVM belongs (which is in a server room).

I needed a 4k switch. I had a nice monitor and a Linux and Windows laptop. I used the Win laptop for gaming and the Linux machine for pretty much everything else.

Everything I tried was garbage:

http://penguindreams.org/blog/4k-uhd-kvm-switches-the-starte...

There are a few in the $500+ range I have not tried, but you are wrong, there are use cases for 4k switching. I mean, do you really want to have two 4k monitors next to each other when you could just have one for half the cost?

Does the 4k monitor have multiple inputs? If it does, then connect both laptops to different input channels and switch inputs on the monitor as needed. Then presumably all you would need is a USB switch for keyboard and mouse input.

I have often wondered why no one offers wireless keyboard/mouse combos with multiple USB fobs and a switch on the keyboard to change which fob it talks to.

https://www.amazon.com/IOGEAR-Computer-Peripheral-Sharing-GU...

Do you really want to have two computers where only one is ever in use, when you could have just one for half the cost?

Your setup is rather peculiar, and wastes a computer. It is usually done with dual-booting or virtualization, while multiple inputs (which I bet your screen has) or HDMI switch could handle it as well (although without handling keyboard/mouse).

KVM's were designed for server environments, sold to server environments, and only take server environments into consideration where VGA+USB will remain the status quo for years to come.

If you're really set on a KVM, take a 4k-capable standalone HDMI switch, a USB "switch" (https://www.amazon.com/IOGEAR-Computer-Peripheral-Sharing-GU..., or a USB KVM), and an arduino to strap it together as a unit. Unless servers do away with VGA (hahah, no), I doubt you'll find a product for an affordable price that does what you want.

Server room KVMs are still going strong. I used them pretty regularly for a small in-store server room.

PSA:

It's about this "KVM":

https://en.wikipedia.org/wiki/KVM_switch

Not this:

https://en.wikipedia.org/wiki/Kernel-based_Virtual_Machine

Thanks! I was wondering why none of the comments made sense.

I opened the link because I was wondering what is it talking about when KVM (the virtual machine) is more popular today than ever. :)

Great, so there won't be confusion anymore! :)

... which makes this a total non-story in my opinion :)

Of course KVM switches will eventually disappear; when was the last time any of us (minus datacenter staff) has been to the server room since the dawn of the PaaS age?

I wouldn't think they are common in data centers either with all the good OOB management tools now available.

I've only used KVM switches on the desktop. Try to think outside your own experiences.

I could never understand why the makers of KVM the virtual machine thing didn't realize they had overloaded an acronym. It isn't as if they weren't likely to have come across KVMs in their work. Were they non-native-English speakers perhaps?

Software devs aren't sysadmins, I imagine most of the people involved here never have been responsible for actual production physical servers before and never had to use one.

Web front end developers sure. Kernel hackers though??

The title of the post should be modified to say "KVM switch".

The main thing that has pissed me off with technologies like DisplayPort and HDMI is that I cannot actually turn my monitors off when I'm not using my computer. With DVI I could turn the monitors off and the desktop and applications would stay where they were. With HDMI/Displayport it considers the monitor disconnected, hence re-adjusts the desktop and moves everything around.

I would love an option that I could switch on and it would "disable HDCP" on these monitors so they could function like they did before. Sure I couldn't play some videos but that's not a great loss for me.

You must have some dodgy hardware. This is definitely not normal, at least with HDMI.

I just tried it with the monitor right in front of me, even unplugging it from the wall didn't cause the laptop to drop the monitor like it would if I had disconnected the HDMI cable.

Sounds perfectly normal to me. I’ve only ever seen the behavior daemin describes.

See my reply to the subthread commentor.

What I had learnt from investigating this issue is:

* It's been a problem since at least 2012

* HDMI cables have an active-low connection pin, therefore it's easier to fool the video card into thinking the monitor is still connected.

* DisplayPort is an active-high signal, therefore if the monitor powers off, it's the same thing as the cable being physically disconnected.

Ideally I would be able to switch the monitor off in such a way that it still maintains a connection, just that the display is off.

Edit: formatting

Is it necessary to you that the display actually be disconnected from power? I use display port on most of my displays and don't have the issue with simply turning off the monitor. But, that's because the monitor doesn't REALLY turn off. It just goes into an ultra low power mode. If that satisfies you, perhaps you need to get a different monitor is all.

Perhaps. I should take a look at the monitor configuration and see if there's a setting I can change. Something like "monitor power button puts monitor to sleep" rather than "monitor power button turns power off to monitor".

I don't turn the power off at the source, i.e. unplug the monitors.

There are two possibilities:

1. Fake GPU software, which I haven't researched at all but probably exists

2. Forcing your graphics chipset to turn a specific head on and leave it on

Here's a Linux-specific example of how to do #2. Since I'm just talking to graphics hardware here, there should be equivalent operations for all of the following for other operating systems.

I don't connect an external display at any point.

First of all, how things look to begin with:

  $ xrandr   
  Screen 0: minimum 320 x 200, current 1024 x 768, maximum 4096 x 4096
  VGA-0 disconnected (normal left inverted right x axis y axis)
  DVI-0 disconnected (normal left inverted right x axis y axis)
  LVDS connected 1024x768+0+0 (normal left inverted right x axis y axis) 0mm x 0mm
     1024x768      60.00*+
     800x600       59.86  
     848x480       59.66  
     720x480       59.71  
     640x480       59.38

Now, `xrandr --output VGA-0 --auto` isn't going to do anything because --auto will tell the GPU to pick the best resolution. No screen is attached so there's no list of resolutions to pick from.

So, I'll generate a generic CVT modeline that sets up the VGA output for 1024x768@60Hz.

  $ cvt 1024 768 60
  # 1024x768 59.92 Hz (CVT 0.79M3) hsync: 47.82 kHz; pclk: 63.50 MHz
  Modeline "1024x768_60.00"   63.50  1024 1072 1176 1328  768 771 775 798 -hsync +

Cool. Now I'll tell X11 about the modeline (note the copy-paste):

  $ xrandr --newmode  "1024x768_60.00"   63.50  1024 1072 1176 1328  768 771 775 798 -hsync +

...And now do the XRandR dance to actually tell the output about the right modeline (this is Xorg-bureaucracy-specific):

  $ xrandr --addmode VGA-0 "1024x768_60.00"

Let's review the xrandr output:

  $ xrandr
  Screen 0: minimum 320 x 200, current 1024 x 768, maximum 4096 x 4096
  VGA-0 disconnected (normal left inverted right x axis y axis)
     1024x768_60.00  59.92  
  (...)

At this point I can go ahead and

  $ xrandr --output VGA-0 --mode "1024x768_60.00"

and my screen displays garbage for a fraction of a second and then figures itself out.

At this point I know the output is enabled (and the glitch didn't mean nothing) and in mirroring mode. I can prove this without a display attached by running xrandr again:

  $ xrandr
  Screen 0: minimum 320 x 200, current 1024 x 768, maximum 4096 x 4096
  VGA-0 disconnected 1024x768+0+0 (normal left inverted right x axis y axis) 0mm x 0mm
     1024x768_60.00  59.92* 
  (...)

Two things: one, the mode has an asterisk next to it, meaning it's active (albeit without a screen attached), and two (the following will make sense in a minute if it doesn't straight away) Screen 0 has a "current" size of 1024x768. (Xorg "screens" represent a collection of video outputs you can move windows between.)

So, let's disable mirroring mode and make this new Schrödinger-pseudo-existent area its own independent surface.

  $ xrandr --output VGA-0 --left-of LVDS

At this point, i3 (my windowmanager) decides to move every single window onto the new offscreen surface area, and I go blind. Isn't minimalist software wonderful? (A new windowmanager is on my todo list; i3 drives me nuts. Anyway.)

Once I sort i3 out and get back to a terminal, xrandr shows something new and interesting:

  $ xrandr
  Screen 0: minimum 320 x 200, current 2048 x 768, maximum 4096 x 4096
  (...)

Oh hey, the Screen size is now 2048x768! Proof that I have two 1024x... outputs configured side-by-side. If I wanted to I could do --above or --below and have a 1024x1536 display.

Here's the technical moral of the story: since approximately the time of the dinosaurs monitors have used EDID to provide autodetection info to graphics chipsets, so forcing manual configuration is pretty deeply buried.

But that's what you need to do; "my screen is weird and needs a custom configuration" and "okay just set these parameters and leave the output head running" are analogous.

If you want to turn your screen off and have the GPU keep the head running, I recommend figuring out how to fish out the modeline your GPU is currently using, and then force the GPU to manually use that modeline. That should disable output autodetection and keep the output permanently active.

Despite the fact that I've demoed Linux output configuration here, you're (ironically) not too likely to have this problem on Linux as Xorg is fairly flexible, and wouldn't normally need these kinds of instructions provided to it under normal circumstances; for Linux it would be more "what's explicitly killing the output and how do I turn it off" as opposed to digging around in driver settings.

Also - I picked VGA for this example, honestly probably for nostalgic reasons. These instructions will work equivalently regardless of output type.

>At this point, i3 (my windowmanager) decides to move every single window onto the new offscreen surface area, and I go blind.

Xmonad can't do stuff like this because it treats each monitor as a separate virtual screen.

If I wanted a bigger monitor I would of just bought a bigger monitor, dammit …

This comment deserves a github gist, and a separate top-level post.

I've never really understood the desire to use the button on the display to turn off monitors - does allowing the monitors to be switched off by the OS not solve basically the same problem ?

I leave my PC running 24/7 pretty much, so at night I want the monitors turned off so it doesn't illuminate the whole living room (which I can see from the bedroom - it is a small apartment).

I used to turn them off with the button on the monitor (when all my monitors were DVI), so if the mouse was moved by accident during the night it wouldn't wake the monitors up. Right now I have to let it go to sleep normally, and then hope there isn't some strange vibration that is big enough to trigger a wake.

Edit: Better language and more info.

Every monitor I've owned since the early 2000s will switch itself off (or at least go into a standby state where the screen is entirely off and there's a tiny blinky LED indicating that it's still powered) after a period of inactivity. I know some monitors never give up waiting for a signal but at this point maybe it'd be simpler to treat yourself to a new monitor?

Edit: Oh, you're talking about things waking your computer up? Maybe turn off 'wake on USB'? Although that's less convenient otherwise. I know when my desktop computer is fully hibernated, mouse motions don't wake it but button presses do. And you can get into this state (assuming Windows 8+) by right clicking on the Windows icon / 'start' button and selecting 'sleep'.

In my experience, Windows 10 seemed to have issues with the "Wake on USB" setting about 6-8 months ago. No matter if I let my computer hibernate or forcibly sleep my computer, it would turn on within 20-30 minutes of me being away. No matter what settings I set, this would happen constantly.

Disabling wake from USB is not a good idea given I have a USB keyboard and mouse.

Essentially this is a convenience thing, it used to work perfectly fine, then they added HDCP to make things less convenient.

Like I said, I'd be fine if the software (well Windows in my case) has a switch to disable HDCP and restore the old functionality of always assuming there's a monitor there.

I came up with a trick for this. Wrote a little script for windows (I'm sure you could do the same thing for linux) that just sends the "sleep" command to the monitors at will.

I can dig it up for you if you want it when I get home.

EDIT: Upon re-read this doesn't really solve your accidental wake up problem. Oops.

You can disable mouse-wakeup, so that the only way to wake from sleep is a keyboard press.

Just leave your mouse upside down.

> I've never really understood the desire to use the button on the display to turn off monitors - does allowing the monitors to be switched off by the OS not solve basically the same problem ?

Right now, my monitor is in my bedroom. If I let the OS shut off the monitor, it still shows a little white pulsing light to show it's in sleep mode. That's really annoying when I'm trying to sleep.

And my monitor has an extremely bright pulsing white LED. I have a piece of orange electrical tape over it, and it is still on the bright side. Dell e2414h if anyone's curious.

A permanent option sometimes required for headless operation: a hardware emulator/terminator such as https://amzn.com/B00FLZXGJ6 ($8.50).

When I worked in the Pentagon, everybody in our office had these to switch between the classified and unclassified computer systems. It was amusing when someone in our large cubicle farm would say, "Hey, check out this email that just came in on the low side," and then two dozen index fingers would reach out and two dozen monitors would blink over to the Windows login screen simultaneously.

G-man here (not the U.S.). Switching between four systems daily. Scroll Lock x2 + n.

Ironically, to me personally KVM [1] is reason #1 for not owning and using the other KVM [2]. I simply do not need an additional physical machine as what I need is available in a VM at no cost with plenty of performance.

[1] https://en.wikipedia.org/wiki/Kernel-based_Virtual_Machine

[2] https://en.wikipedia.org/wiki/KVM_switch

I'd guess the comment about HDCP is onto something. To build a KVM switch that works with modern displays without weird errors, it probably has to be an active HDMI/HDCP endpoint on its own - which means a vendor would either need to get its own sets of keys and do additional work to enforce all DRM restrictions - or they'd have to use leaked keys and prepare to get sued in an instant.

Even without HDCP, I remember another article explaining that it's already difficult enough to get two arbitrary HDMI devices to talk to each other because HDMI allows so many different modes of operation. I imagine the situation would get even worse if you had to negotiate common protocols for three endpoints at the same time.

What if it physically unplugged the HDMI connection, and reconnected it to the other device? It wouldn't be quite as seamless since the initialization process would be repeated, but it would work.

Not even physically, but just redirecting the raw signal to another device should work. Maybe artificially break it for a second or so to better simulate plugging out and back in. I can't see how the actual endpoints could tell a difference between that and manually un- and re-plugging into a different device.

That's what I meant. Electrically, it's identically to unplugging from one device and plugging into the other.

This is doable but unappealing as the OS will see the display is gone, and relayout the desktop environment every time you switch. Depending on your OS, and your personal concerns about all your environment behaviors, you might not care. I know I would be very annoyed by this.

You'd need the HDCP master key, but you need not be a full-featured endpoint. See https://www.bunniestudios.com/blog/?p=2117 for an example of a device that mixes an incoming HDMI video with a (non-HDMI) overlay video. The device is sold in the US, so it must be considered legal.

HDCP seems to work absolutely fine with the plethora of HDMI switches available on the market right now, so that can't be the issue.

My monitor has a KVM built in...

http://www.dell.com/en-us/work/shop/dell-34-ultrasharp-curve...

It has two USB ports that can go to two different computers, and then in the settings you can set it up so that when you switch from one input to another it also switches what the USB devices are connected to. There are 4 USB ports total on the monitor, and they switch from one device to the other.

It's great, instead of being a separate box, it's built into the monitor. It's neater.

well this monitor also can daisy chain another monitor. which is awesome per se, and actually the daisy chained monitor gets switched aswell. (http://en.community.dell.com/support-forums/peripherals/f/35...)

but 800 bucks for a monitor :/ I've always looked at these dell monitors. but I never use a external monitor when I'm at home (laptop). But at work my boss would never pay for any of those.

I work from home. I bought it for myself because I like having the extra real estate. Best purchase I have made in a long time.

Also, I couldn't imagine two of these next to each other, 34" is very large!

Synergy (https://symless.com/synergy) may have had something to do with it.

Yeah, I can see how it might not work for everyone but for my needs it's pretty well suited and become part of my everyday workflow.

Yep, that is what I use for the same usecase. There is really very little point to a hardware KVM at this point.

The remaining use case is when you want to share a monitor along with the keyboard and mouse.

I cannot live without this software. I'm using it to control several computers with a wall of screens. My T460p is the server.

FYI synergy is free and oss[1] but I would still recommend paying for it because it's priced fairly and I want to support the development behind it. I think synergy is text-book "how to make money with oss."

[1] https://github.com/symless/synergy-core

I, too, tried in vain to buy a decent quality 4k KVM. Oddly the video worked better than the USB mouse/keyboard. (?)

Having failed the KVM route, I, too became a Synergy fan. I don't think Synergy caused the decline of KVM, but it synergy's success is certainly correlated. :)

These days, I believe the monitor is becoming the KVM. High-end Dell laptops have multiple inputs as well as USB slots for peripherals, and even a handful of USB type B to send back downstream.

Mine doesn't, unfortunately, so for my two computers plugged into my Dell monitor, I have the keyboard and mouse plugged into a cheap USB switch, while simply cycling through the inputs on the monitor.

It's two clicks instead of one with a traditional KVM, but I don't have $150 to blow on a KVM that has something better than a VGA slot...

With VGA, it was simple: duplicate the EDID to both inputs, switch the output wires, done. AFAIK, with DisplayPort it's not so simple, since there is a negotiation between the computer and the display, including link training (see https://lwn.net/Articles/736011/ for instance). IIRC, HDMI also has some bidirectional communication. To work well, a KVM would have to act as a display to the computers, and as a computer to the display, instead of just passively switching some wires.

I read this "article", and a few comments below it - and the replies, and I am just wondering:

Who is this "Stanislav", and why are they being so grating on people replying?

I mean, I can understand disagreement and correction (ie, the whole "please re-read the article, your question was answered" or "the article already addresses this", etc) - but do they have be such an arrogant a about it?

It basically turned me off from reading further - or exploring the rest of the blog.

I use two machines as part of normal operation, my monitors have multiple inputs which can be switched with one button

and with synergy the mouse/keyboard can move across

no need for a KVM

This! I was a big user of KVMs for many years, but haven't used one for the past ten(ish) years after discovering Synergy. I don't switch monitor inputs, but have had a bunch of them connected to various computers, and being able to switch with no effort at all is great.

Wendell, from the YouTube channel Level1techs, has worked with a Chinese(?) OEM on custom firmware for Displayport KVMs and is selling them:

https://twitter.com/tekwendell/status/933441036138971137?s=1...

Still more expensive than a 4k monitor, but heading in the right direction.

One of the big remaining users of display switching is the military. When something important is happening on one screen, others need to look at that screen. The military wants switching so they don't get everybody jammed around the screen with the action and abandoning their own stations. Here's the current incarnation of NORAD HQ in Cheyenne Mountain.[1]

Such switching goes all the way back to NASA during Apollo. Their system put all the screens on a video cable as cable TV signals. Each console, and other locations around Mission Control and Launch Control, all had channel selectors and could watch any display. Using different technology, that capability was maintained in later control centers.

[1] https://en.wikipedia.org/wiki/Tactical_operations_center#/me...

If the military needs to spend $1000 on a KVM switch, they won't bat an eye. That's not true for most of the rest of us.

I've seen such kind of "control rooms" where everything was running on VMs anyway and the individual computers were just remote desktops. It's like they say, when it gets too difficult, just add a layer of abstraction. They just went up to software.

I replaced them with standalone USB & HDMI switches. I connect 5 PCs on my desk (main PC, Deep Learning rig, NAS, 2x NUC) first to two 4-port USB switches, then have one 2-port switch that switches between these two 4-port switches, and then a large HDMI switch. It works fine; each of those switches was <$5.

What I miss about KVM switch is the ability to switch to another system using a keyboard hotkey, such as double-pressing the "Scroll Lock" key on the keyboard. Most high-end KVM had this feature, so that I could use a hotkey to quickly switch to different system. Once anyone gets used to this, there's no going back to rotary dials or slow physical push buttons, etc. I had 3 or 4 systems connected to KVM and nothing was faster.

I've tried using modern HDMI/monitor switchbox/splitter and USB switches, but they require awkward physical reaching for the buttons, as I have to reach out for a button or two, every time I need to switch to another system. The old KVM did this in split second and I didn't even have to think about it, sort of like Vim once you get used to the keys. I was able to work so much faster, switching from one system to another, whenever I needed to. Now, I have to think about it before switching, not that I'm lazy, but I don't want to exert so much energy just to switch to another system.

Edit: I also miss PS/2 and VGA, for their ability to switch quickly as well. USB devices and HDMI monitors take awhile to switch.

I use this one for my 40" 4k monitor, and it works great : https://www.amazon.com/gp/product/B01BSNSOPU/

Oh boy, the price! The article is lamenting the expense of 'modern' KVM switches.

For reference, a switch with 'legacy' USB 2.0/VGA connectors that powers my ageing home PCs is this model:

https://www.ebay.com.au/itm/All-in-one-Mini-2-Ports-KVM-Manu...

That's the second endorsement of an IOGEAR DisplayPort solution I've seen so far. The full product line is here: https://www.iogear.com/solutions/kvm

$480 to solve the dual head 4K DisplayPort problem, and it works, apparently.

Now, is that a reasonable price? I have to say I think so. This is not a trivial problem and this equipment isn't being manufactured by a dozen companies and flying off Best Buy shelves. Given the low demand and difficult requirements I think the price is reasonable. If your use case involves smoothly switching between 2-4 high resolution dual head systems $480-$680 probably amounts to something in the range of 5-20% of the cost of the entire system.

I don't buy the argument that this cost makes such a KVM pointless because one can simply buy more display(s) and input devices instead: that takes a lot of space and results in a lot of underutilized hardware.

I think it's a two pronged demise

* Current display (DP, HDMP) and peripheral(USB) technology is complex, and probably is going to be expensive to switch in the fly without being unreliable.

* Virtual Machines and Remote Desktop solutions are good enough. Sure they can be a bit laggy sometimes, but good enough for most tasks.

So i guess that KVM switches are going to be more expensive to make, and there are less customers who really need them...

I think the use cases are disappearing. Using a KVM means working near many noisy computers. I do not know the location of most of the machines I am using remotely from my dumb windows PC. Poor guys that must use a KVM.

I still would like a personal one, I get to work from home frequently, and have a personal desktop, personal laptop and work laptop that'd be great to switch between without having to keep changing plugs (as it is, I velcro the HDMI, mouse, keyboard and speaker jack together to keep it easy to handle).

We are still using it in our testing environment - 8+ servers. We do a lot of pre-OS tests so KVM is the optimal solution - faster and more reliable than IPMI/etc.

There is a project to build an open source KVM-over-IP Rapsberry pi based.

rasky:

https://www.nexlab.net/product/rasky/

I have contributed to it but a lot of time without news. Maybe contribute to rasky was risky.

You can get a USB switch and a 4K HDMI switch for about $40.

I don't currently have a 4K HDMI switch, but my HD switch works just fine. Sure you have two buttons to push (USB and HDMI), but it works and it's cheap.

Yep, I got this for my 4k monitor: https://www.amazon.com/gp/product/B071CJWSPF/

And this for my usb stuff: https://www.amazon.com/Plugable-One-Button-Swapping-Between-...

It's under $25 total and both work perfectly well.

One interesting use-case is having virtual machines with dedicated ports.

Haha, those were the days!

I had one for my dual desktop setup back in the days.

Had a Windows and a Linux machine, but was too poor for second mouse/keyboard/monitor. The Linux machine was some old PC I built from spare parts.

I already had a switch that let me change the PC via ctrl+esc on the keyboard.

At my first job as a PC technician we still had the old ones with physical switches.

The ones that used VGA ports for video often used to degrade the image quality in comparison to a direct connection.

Frankly there are entire categories of desktop computer peripherals that get a deserved reputation for being flakey junk. USB hubs used to be this way, but presumably they've finally got those right. Inkjet printers too.

    USB hubs used to be this way, but presumably they've finally got those right.

Not necessarily. I purchased a 10-port hub just a couple years back that was internally three 4-port hubs chained serially. The OSX USB stack couldn't see devices connected to the last four ports...

This does not surprise me at all. I have a 4k screen and just thinking about all the fiddly bits between HDMI, HDCP and Freesync/Gsync between the computer and the screen makes it quite difficult to build a switch. VGA was primitive, but simple to switch, those times are gone. My computers usually make a fuss if I just switch off the screen and back on, because both sides seem to have some active signaling going on and way too much logic.

Ps: USB switches for the KM parts are rather simple though and there is a healthy offering (+ you can build this yourself with some extension cords and a simple USB switch). Also, more and more of my devices move wireless (e.g. audio).

As far as I am aware from the last time I tried to find such a thing the packet-based interactive nature of modern display standards prevents the cheap implementation of these devices, as opposed to the non-interactive, older standards.

This tone of this article is pretentious and just plain douchey. The author is largely assuming that their small niche (and petty) use case is what everyone who uses a KVM would want (i.e. using a 4k monitor). The single largest use for KVM is for servers, who the hell is wasting money on a 4k monitor for their servers? There are plenty of reliable and affordable KVM's that the author deems as old relics that work just fine for the large majority of consumers.

> This tone of this article is pretentious and just plain douchey.

I got that same feeling. First a "What?! KVMs are no longer sold at all?!" followed by "wait, no, he's lying - KVMs are still sold, they just don't meet his standards." It's clickbait.

The rise of more powerful graphics cards with the ports and ability to run multiple monitors made this something that ever fewer desktop users need that did before. You can still get good KVMs in the enterprise/hardware market, they are common in server rooms, but even then less so than the old days as most firmware allows remote management.

I don't understand what this has to do with graphics cards. One is about connecting multiple monitors to one computer, the other about connection one monitor to multiple computers.

For example, my 3 monitors each have 4 different inputs. My GPUs have at least 3 outputs. So you can connect each GPU to each monitor, then you need a way to switch between different computers at once or separately.

Are there any not-too-expensive, modern (as in DisplayPort-friendly and USB), hardware KVMs available these days?

I started looking into a solution and I ended up shelling out money for Synergy, which was cheaper than the HDMI-or-Displayport-I-forgot KVMs I saw on Amazon, which looked like they were in the 2,300+ range (and with really poor reviews).

I bought a USB2.0/VGA switch online for < $10. I had a number of vga and USB type B (aka printer) cables lying in boxes that are now put to good use!

A windows laptop and a linux NUC, 5 USB peripherals shared.

4K and all digital HDMI might spell the death of it but I'd be needing new PC(s) to drive such a display in any case.

This URL starts off with the condition that the KVM needs to be less expensive than adding another display and keyboard, and operate at 4K. So this is targeted towards the home environment? I'd guess that RDP/VNC and the like have pretty much done away with the KVM in that environment.

I am so happy I don't use KVMs any more. Instead, monitors with multiple inputs, and a simple USB switch for the keyboard.Then I don't have to spend $200 when I upgrade monitors, just so I can switch whatever new resolution I moved to.

I've looked into doing this as well. Have you found monitors that a.) switch inputs without many seconds of delay b.) switch using keyboard input as opposed to banging on balky switches on two displays? My expectation is one button and no delay, not switching the input devices and then switch two monitors one after the other, because I expect to switch frequently. Also, the buttons on a lot of monitors aren't really intended to be used terribly hard; they wear out. Maybe capacitive buttons solve this, I guess, but those can be flaky as well for other reasons.

no, unfortunately all the monitors I use are slow (3-5 seconds) and the keyboard switch takes 2-3 seconds.

I think this is because people perceive multiple monitors as giving a better overview, regardless of switches being available at a lower cost, which has caused the market for switches to become a niche market.

Jesus that author has a real attitude problem.

All of our meeting rooms have computers with KVMs in case someone wants to use their laptop on the big screen.

Or they can just remote in. It's kind of dumb.

How do you "just remote in" with no (working) OS?

'Smart' displays?

Where I worked they wall-mounted large screen TVs. Windows 10 casted to them over wifi just fine.

I used to have a KVM to switch between my many boxes with many OSes. Now with virtualization, I have box with a few virtual machines. Now if I want to use windows, debian and freebsd, I don't have to switch between 3 physical boxes, I just have to switch virtual machine windows on one box.

Reply on news.ycombinator.com