Recommendation for 16+ Port KVM Switch

Once upon a time, we had all of our rackmounted server room computers hooked
into multiple Belkin eight-port KVM switches, and life was good (or at least
tolerable). The customizable menu with names on it would periodically
reset itself and lose all values, but at least we could switch from machine
to machine. We have "upgraded" to the latest Belkin 16 port model, and
it has been pure hell. The device is constantly having problem with
keyboard or mouse (sometimes both) on at least 20% of the connected
machines. We have tried new cables, and it doesn't seem to make any
difference. Belkin replaced the switch, and the problem gets worse. I'm
looking for recommendations for a new vendor, possibly a new architecture,
for a large scale KVM solution for a server room.

My basic requirements are as follows:

1) Up to 32 computers, each of these controllable from two to four stations.

2) Any station can control any computer.

3) We should be able to run long distance, at least 100 feet, from the
keyboard, mouse, and monitor to the actual server.

4) Should allow us to identify easily each machine using a name we can
assign to it and then read off the KVM switch on screen menu.

5) Should be incredibly *reliable*.

Perhaps a better solution would be one that puts the port selector right at
the keyboard and mouse and monitor, then runs a multiplexed KVM signal over
RJ-45 to a back end device that then attaches to each server. I'm open to
any and all ideas on how to best accomplish the mission here.

--
Will

In article <CP2dncMFOIl->, "Will"
<westes-> wrote:

> Once upon a time, we had all of our rackmounted server room computers hooked
> into multiple Belkin eight-port KVM switches, and life was good (or at least
> tolerable). The customizable menu with names on it would periodically
> reset itself and lose all values, but at least we could switch from machine
> to machine. We have "upgraded" to the latest Belkin 16 port model, and
> it has been pure hell. The device is constantly having problem with
> keyboard or mouse (sometimes both) on at least 20% of the connected
> machines. We have tried new cables, and it doesn't seem to make any
> difference. Belkin replaced the switch, and the problem gets worse. I'm
> looking for recommendations for a new vendor, possibly a new architecture,
> for a large scale KVM solution for a server room.
>
> My basic requirements are as follows:
>
> 1) Up to 32 computers, each of these controllable from two to four stations.
>
> 2) Any station can control any computer.
>
> 3) We should be able to run long distance, at least 100 feet, from the
> keyboard, mouse, and monitor to the actual server.
>
> 4) Should allow us to identify easily each machine using a name we can
> assign to it and then read off the KVM switch on screen menu.
>
> 5) Should be incredibly *reliable*.
>
> Perhaps a better solution would be one that puts the port selector right at
> the keyboard and mouse and monitor, then runs a multiplexed KVM signal over
> RJ-45 to a back end device that then attaches to each server. I'm open to
> any and all ideas on how to best accomplish the mission here.

There is a box here (KM0432) that handles 32 computers, with four
stations connecting to the 32 computers. A converter box is connected
locally to each computer, and the backhaul can be up to 150 meters
(at 1024x768 res).

http://www.aten.com/products/product...erid=subClass1

The KA9120 is an example of a PC compatible converter for the KVM.

http://www.aten.com/products/product...60414155726007

And judging by the hits in a search engine, there are other similar ones.

http://www.altavista.com/web/results?q=32+port+kvm

$6000-$7000 ought to handle it, depending on how greedy the supply
chain is.

Maybe you could arrange to get an eval unit, for some testing.
Otherwise, I doubt you'll get a lot of feedback on a big ticket
item like this.

Paul

"Paul" <> wrote in message
news:nospam-3107062248390001@192.168.1.178...
> There is a box here (KM0432) that handles 32 computers, with four
> stations connecting to the 32 computers. A converter box is connected
> locally to each computer, and the backhaul can be up to 150 meters
> (at 1024x768 res).
>
>
http://www.aten.com/products/product...erid=subClass1
>
> The KA9120 is an example of a PC compatible converter for the KVM.
>
>
http://www.aten.com/products/product...60414155726007
>
> And judging by the hits in a search engine, there are other similar ones.
>
> http://www.altavista.com/web/results?q=32+port+kvm
>
> $6000-$7000 ought to handle it, depending on how greedy the supply
> chain is.

At $6000, ATEN is about 1000% more expensive than what I have seen similar
solutions go for. Seems like a lot more than what we would need to spend.

The KA9120 in an intriguing idea. If I'm understanding this correctly,
the KA9120 presents a DB15 video, and a PS/2 mouse and keyboard, at either
end, joined in the middle by RJ-45? Do they have an active amplifier of
some kind in this box to travel the 150M? What distance can they carry
1600x1200, which is our standard for some of the servers we need to control?

--
Will

An additional requirement I forgot to add:

6) We need 1600x1200 resolution for some of the servers.

--
Will


"Will" <westes-> wrote in message
news:CP2dncMFOIl-...
> Once upon a time, we had all of our rackmounted server room computers
hooked
> into multiple Belkin eight-port KVM switches, and life was good (or at
least
> tolerable). The customizable menu with names on it would periodically
> reset itself and lose all values, but at least we could switch from
machine
> to machine. We have "upgraded" to the latest Belkin 16 port model, and
> it has been pure hell. The device is constantly having problem with
> keyboard or mouse (sometimes both) on at least 20% of the connected
> machines. We have tried new cables, and it doesn't seem to make any
> difference. Belkin replaced the switch, and the problem gets worse. I'm
> looking for recommendations for a new vendor, possibly a new architecture,
> for a large scale KVM solution for a server room.
>
> My basic requirements are as follows:
>
> 1) Up to 32 computers, each of these controllable from two to four
stations.
>
> 2) Any station can control any computer.
>
> 3) We should be able to run long distance, at least 100 feet, from the
> keyboard, mouse, and monitor to the actual server.
>
> 4) Should allow us to identify easily each machine using a name we can
> assign to it and then read off the KVM switch on screen menu.
>
> 5) Should be incredibly *reliable*.
>
> Perhaps a better solution would be one that puts the port selector right
at
> the keyboard and mouse and monitor, then runs a multiplexed KVM signal
over
> RJ-45 to a back end device that then attaches to each server. I'm open
to
> any and all ideas on how to best accomplish the mission here.
>
> --
> Will
>
>

In article <_>, "Will"
<westes-> wrote:

>
> At $6000, ATEN is about 1000% more expensive than what I have seen similar
> solutions go for. Seems like a lot more than what we would need to spend.
>
> The KA9120 in an intriguing idea. If I'm understanding this correctly,
> the KA9120 presents a DB15 video, and a PS/2 mouse and keyboard, at either
> end, joined in the middle by RJ-45? Do they have an active amplifier of
> some kind in this box to travel the 150M? What distance can they carry
> 1600x1200, which is our standard for some of the servers we need to control?

The KA9120 is ~$100 per unit, and if you used 32 of them, that is
$3200. So it is not a cheap option. It converts the connectors
at the computer end, into an RJ-45 based signal. The "Matrix KVM"
at the other end, has RJ-45 connectors ready to receive those
signals. So the KA9120 is not a generic remoting solution, it is
a specific extension for their switch. The fact that there are
different flavors of KA9120-type devices, allows connecting
computers other than PCs.

If you wanted a generic remoting solution, I'm sure there are
some out there.

As for resolution, there is no table of distance versus resolution,
but is says 1920x1440 at short distance. So 1600x1200 is possible,
but not with the full 150M cable limit. Basically, that tells you
the method is analog, with some bandwidth versus distance relationship.
If you find an extender with fiber optic interface on it, that
can go longer than 150M, but with a much higher price.

The Aten is simply the first hit from the search engine I looked
at. There looks to be plenty of "copy-cat" action, so perhaps
you'll find something with a more reasonable price. But since
the unit is 32 ports, with flexible 4 stations at the other end,
this is a unit intended for big corporate users, with big
corporate budgets. You've got to expect some gouging, if you
play in a market like that. They won't be selling a million
of these a year, and the staff gotta eat :-)

Paul

Will wrote:
> Once upon a time, we had all of our rackmounted server room computers hooked
> into multiple Belkin eight-port KVM switches, and life was good (or at least
> tolerable). The customizable menu with names on it would periodically
> reset itself and lose all values, but at least we could switch from machine
> to machine. We have "upgraded" to the latest Belkin 16 port model, and
> it has been pure hell. The device is constantly having problem with
> keyboard or mouse (sometimes both) on at least 20% of the connected
> machines. We have tried new cables, and it doesn't seem to make any
> difference. Belkin replaced the switch, and the problem gets worse. I'm
> looking for recommendations for a new vendor, possibly a new architecture,
> for a large scale KVM solution for a server room.

Try different mouse and keyboard. I used to get this on my Lynksys KVM
too when I was using wireless keyboard and mouse but since switching to
standard Logitech PS2 keyboard and Logitech MX518 mouse using PS2
adapter I no longer have any issues.

Will wrote:
>
> My basic requirements are as follows:
>
> 1) Up to 32 computers, each of these controllable from two to four stations.
>
> 2) Any station can control any computer.
>
> 3) We should be able to run long distance, at least 100 feet, from the
> keyboard, mouse, and monitor to the actual server.
>
> 4) Should allow us to identify easily each machine using a name we can
> assign to it and then read off the KVM switch on screen menu.
>
> 5) Should be incredibly *reliable*.

What about RDP or VNC?

--
Mike Smith