FIOS: Lost Connectivity, Replaced ONT, Now Only Verizon Router Works?

Per Char Jackson:
One testing approach is to connect two Gigabit-capable PCs....

Switch arrived today, been working that approach.

One more question: Assuming a gigabit switch with all devices
connected to it capable of gigabit except one device that only does
10/100; is that device dragging everybody else's speed down?

If so, would replacing it with another gigabit switch and then plugging
the 10/100 device into that second gigabit switch mitigate the
situation?
--
Pete Cresswell

Per Char Jackson:
Those are strange numbers. They're faster than 100, so you obviously have
Gigabit capable network devices under test, but they're significantly less
than 1000 so something's not quite right. My guess is network congestion.
More below.

I think your guess is correct.

For ease of testing, I combined all my IP cams into one little switch
and then plugged that switch into the main switch.

After disconnecting/re-connecting the little switch over-and-over it
started to dawn on me that it was the traffic from the IP cams that was
pulling me down to those 200 Mbps speeds. Without the cams speeds were
around 450.... replicable over-and-over....

Makes me think of one of the System Programmers where I used to work
back in the mainframe days: he used to get all bent out of shape because
all those people logged in were dragging down his performance
numbers....

But, speaking of numbers, my IP cam server claims that the total
bandwidth for all seven of those IP cams is only 5.8 Mbps.
OTOH maybe I am misreading it and there is more going on than just it's
numbers per camera....
--
Pete Cresswell

On Fri, 05 Jun 2015 21:21:47 -0400, "(PeteCresswell)" wrote:

Per Char Jackson:
One testing approach is to connect two Gigabit-capable PCs....

Switch arrived today, been working that approach.

One more question: Assuming a gigabit switch with all devices
connected to it capable of gigabit except one device that only does
10/100; is that device dragging everybody else's speed down?

No, not quite. When any two devices talk to each other through a switch,
they should talk at the lowest common speed that both of them support. So
two gig devices should talk gig, while two 10/100 devices should talk 100,
and a gig device and 10/100 device should talk 100.

Multiple conversations can be going on simultaneously through the switch,
and the slower speed of one conversation doesn't affect the faster speed of
another.

If so, would replacing it with another gigabit switch and then plugging
the 10/100 device into that second gigabit switch mitigate the
situation?

Not necessary.

--

Char Jackson

On Fri, 05 Jun 2015 21:54:20 -0400, "(PeteCresswell)" wrote:

Per Char Jackson:
Those are strange numbers. They're faster than 100, so you obviously have
Gigabit capable network devices under test, but they're significantly less
than 1000 so something's not quite right. My guess is network congestion.
More below.

I think your guess is correct.

For ease of testing, I combined all my IP cams into one little switch
and then plugged that switch into the main switch.

After disconnecting/re-connecting the little switch over-and-over it
started to dawn on me that it was the traffic from the IP cams that was
pulling me down to those 200 Mbps speeds. Without the cams speeds were
around 450.... replicable over-and-over....

Years ago, I remember having a no name brand NIC that supposedly did gig,
but I was lucky to get 450 Mbps out of it. Yours reminds me of that.

Makes me think of one of the System Programmers where I used to work
back in the mainframe days: he used to get all bent out of shape because
all those people logged in were dragging down his performance
numbers....

I resemble that remark. :-)

But, speaking of numbers, my IP cam server claims that the total
bandwidth for all seven of those IP cams is only 5.8 Mbps.
OTOH maybe I am misreading it and there is more going on than just it's
numbers per camera....

There's a formula for calculating what the required bandwidth should be, but
I'm not enough of a geek to remember the details. I only remember that it's
based on the resolution times the frame rate, and then I forgot.

--

Char Jackson

Char Jackson wrote:


Years ago, I remember having a no name brand NIC that supposedly did gig,
but I was lucky to get 450 Mbps out of it. Yours reminds me of that.

RealTek makes a PCI bus GbE chip like that.

I have four PCI cards like that, I can sell you
real cheap. I'm sorry I bought them. They're not
even fit to be used as door-stops.

To get that RealTek to run at full link rate, by extrapolation,
it would take a 4GHz Core2 processor to make it happen.
So something like a 4790K combined with one of those
stinkers, you'd just barely make it to full link rate.
The chip has some sort of "excess interrupt" problem.
About five interrupts per packet. Not only is it slow,
it chews up CPU as well when you use it.

Paul

Per Stormin' Norman:

Have you tried adjusting the frame rate of each camera? Have you previously
mentioned the make and model or at least the resolution and frame rate of each
device, if so, I didn't see it.

If you are using the cameras for security and surveillance, in typical
situations setting the frame rate to 1 or 2 fps should be more than adequate.
You might also take a look at your resolution, the higher the setting the
greater the amount of data per frame.

I am just trying to get a handle on the "What" and "Why" of my network's
speeds.... I can live with 200 Mbps, no problem... just want some idea
of what is going on.

My cams are all set up to be bandwidth-intensive - set to 30 FPS (mostly
HikVision DS-2CD2032-I's and one higher-end Sony. My cam server
defines each camera twice:

- In one definition, a cam takes contiguous 5-minute Clips back-to-back
24-7.

- In the other definition, a cam takes no Clips, just motion-sensor
Alert .JPGs. These occur once per second while motion is being
sensed. Frankly, for my money, I could totally live without clips
if the use was only security. The one-second .JPGs are plenty.
And they have the additional attraction of being quickly FTB-able
so if something happens on-site to disable the connection at least
some of the .JPGs will get through - as opposed to Clips that take
so long to transmit and which are not playable unless the entire
file is present.

I guess the cams serve some security purpose, but their main purpose for
me has been as learning vehicles and entertainment.

- Learning vehicles because the owner of the windsurfing shop where I
sail wanted an IP cam so people could check on conditions and not call
him every five minutes on windy days. But he was behind a 44k DSL
connection.

It turned out to be quite a little adventure for me - knowing nothing
at the start (and still not knowing much...but, at least it
works...)... http://ExtremeSurfCam.DynDNS.org.

- Entertainment because I use the cams at home mainly to see what's
going down in the back yard at night. Foxes, 'possums, raccoons,
neighbors' cats, and the occasional escapee dog.... they all interact
and it's quite a show sometimes.
e.g. https://www.youtube.com/watch?v=E8N3xyJo5pU,
https://www.youtube.com/watch?v=YXrItm_bmjw

Haven't seen any coyotes yet, but I think it is just a matter of
time. Guy about 3/4 mile away who keeps chickens locks them up now
since coming face-to-face with a coyote in his driveway.
--
Pete Cresswell

On Fri, 05 Jun 2015 23:55:48 -0500, Char Jackson wrote:

On Fri, 05 Jun 2015 21:21:47 -0400, "(PeteCresswell)" wrote:

Per Char Jackson:
One testing approach is to connect two Gigabit-capable PCs....

Switch arrived today, been working that approach.

One more question: Assuming a gigabit switch with all devices
connected to it capable of gigabit except one device that only does
10/100; is that device dragging everybody else's speed down?

No, not quite. When any two devices talk to each other through a switch,
they should talk at the lowest common speed that both of them support.

Correcting myself, they talk at the _highest_ common speed, not the lowest.


So
two gig devices should talk gig, while two 10/100 devices should talk 100,
and a gig device and 10/100 device should talk 100.

Multiple conversations can be going on simultaneously through the switch,
and the slower speed of one conversation doesn't affect the faster speed of
another.

If so, would replacing it with another gigabit switch and then plugging
the 10/100 device into that second gigabit switch mitigate the
situation?

Not necessary.

--

Char Jackson