What is your favorite method to troubleshoot high-bandwidth/high latency networks?

Gurus,

What is your favorite method to troubleshoot high-bandwidth/high latency
networks? I know about the ping and telnet tests you can do, aside from
getting on the phone with the ISP, is there anything one can do from the
internal customer's side of the house in order to find the "smoking gun"?

--
Spin

"Spin" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...

> What is your favorite method to troubleshoot high-bandwidth/high latency
> networks? I know about the ping and telnet tests you can do, aside from
> getting on the phone with the ISP, is there anything one can do from the
> internal customer's side of the house in order to find the "smoking gun"?

I always look at the network infrastructure design,...the "topology" if you
want to call it that. Most of the time,..if they are designed correctly
they work correctly. Most network problems that I run across are caused by
the Admins and their bad designs. Of course the LAN and the WAN are two
different things and have to be investigated separately.


--
Phillip Windell
www.wandtv.com

The views expressed, are my own and not those of my employer, or Microsoft,
or anyone else associated with me, including my cats.
-----------------------------------------------------

"Phillip Windell" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> I always look at the network infrastructure design,...the "topology" if
> you want to call it that. Most of the time,..if they are designed
> correctly they work correctly. Most network problems that I run across
> are caused by the Admins and their bad designs. Of course the LAN and the
> WAN are two different things and have to be investigated separately.

It's difficult to get Verizon, AT&T and other major ISP players to divulge
their network infrastructure design and the type of equipment they use,
though not impossible if your considered a top tier customer. Getting some
engineer on the phone to help pinpoint an SLA problem is another story. I'm
so tired of spending night son the phone with service providers only to find
out the problem was caused by an admin or bad design. Happens a lot.
That's why I'm looking to develop a little checklist of "smoking guns" to
look for, a or at least put something together based on responses to this
thread in this NG.

"Spin" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> "Phillip Windell" <(E-Mail Removed)> wrote in message
> news:(E-Mail Removed)...
>> I always look at the network infrastructure design,...the "topology" if
>> you want to call it that. Most of the time,..if they are designed
>> correctly they work correctly. Most network problems that I run across
>> are caused by the Admins and their bad designs. Of course the LAN and
>> the WAN are two different things and have to be investigated separately.
>
> It's difficult to get Verizon, AT&T and other major ISP players to divulge
> their network infrastructure design and the type of equipment they use,
> though not impossible if your considered a top tier customer. Getting
> some engineer on the phone to help pinpoint an SLA problem is another
> story. I'm

Yea,..that's why we use a local provider that I can hop in the car and drive
over there and grab them by the shirt if I have to.

> so tired of spending night son the phone with service providers only to
> find out the problem was caused by an admin or bad design. Happens a lot.
> That's why I'm looking to develop a little checklist of "smoking guns" to
> look for, a or at least put something together based on responses to this
> thread in this NG.

Well,...it's like I said before. The Internet performance and the LAN
performance have nothing to do with each other. The LAN can run perfectly
fine without the Internet,...like they did before the Interent ever came
along. By the same token the Internet connection can run perfectly fine
without the LAN, like it does with a home user with one machine and no LAN.

So you have to determine what it really is you are looking for. By your
post,...I cannot tell what it is. Is the LAN slow? Is the Internet
slow?,...they are not the same thing.

--
Phillip Windell
www.wandtv.com

The views expressed, are my own and not those of my employer, or Microsoft,
or anyone else associated with me, including my cats.
-----------------------------------------------------

"Phillip Windell" <(E-Mail Removed)> wrote in message
news:%(E-Mail Removed)...
> Well,...it's like I said before. The Internet performance and the LAN
> performance have nothing to do with each other. The LAN can run perfectly
> fine without the Internet,...like they did before the Interent ever came
> along. By the same token the Internet connection can run perfectly fine
> without the LAN, like it does with a home user with one machine and no
> LAN.
>
> So you have to determine what it really is you are looking for. By your
> post,...I cannot tell what it is. Is the LAN slow? Is the Internet
> slow?,...they are not the same thing.

Well, consider the case of a dedicated site-to-site VPN, which is an
Internet technology designed to make remote networks "look" local. In this
case, we have a LAN which is really a WAN but made to look like a LAN.
These are the case for which I speak.

"Spin" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...

> Well, consider the case of a dedicated site-to-site VPN, which is an
> Internet technology designed to make remote networks "look" local. In
> this case, we have a LAN which is really a WAN but made to look like a
> LAN. These are the case for which I speak.

In that case it is the Internet. The VPN Tunnel is not "real" it is an
imaginary thing,...a "logical" thing. So they are just "packets" going over
the Internet,..what is ecapsulated within them is not relevant. VPN by
nature is slower than regular Internet traffic for two reasons:

1. The required encapsulation and encryption has a cost in performance

2. VPN runs Synchronous (same speed both ways), so if the Internet
connection is Asynchronous (fast download, slow upload) then it is going to
run at the slower "upload" speed of the connection minus the protocol
overhead (mentioned in #1)and minus whatever bandwidth is already being used
up by other things.

So,...do some speed tests to see what you are dealing with. Anything less
that about 1,200 mbps is going to perform like crap (IMO).

One of the many speed test sites is http://www.speakeasy.net/speedtest

The next important thing to remember is that there are *two* Internet
connections involved with VPN,...one at each end,...it only takes one of
them to screw it up. It is only going to run at the slowest speed of the
slowest connection, minus the other bandwidth eating factors mentioned
above.

--
Phillip Windell
www.wandtv.com

The views expressed, are my own and not those of my employer, or Microsoft,
or anyone else associated with me, including my cats.
-----------------------------------------------------

How would I test the speed between two buildings which are in two
geographically separate cities in my site-to-site VPN network?

"Phillip Windell" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> "Spin" <(E-Mail Removed)> wrote in message
> news:(E-Mail Removed)...
>
>> Well, consider the case of a dedicated site-to-site VPN, which is an
>> Internet technology designed to make remote networks "look" local. In
>> this case, we have a LAN which is really a WAN but made to look like a
>> LAN. These are the case for which I speak.
>
> In that case it is the Internet. The VPN Tunnel is not "real" it is an
> imaginary thing,...a "logical" thing. So they are just "packets" going
> over the Internet,..what is ecapsulated within them is not relevant. VPN
> by nature is slower than regular Internet traffic for two reasons:
>
> 1. The required encapsulation and encryption has a cost in performance
>
> 2. VPN runs Synchronous (same speed both ways), so if the Internet
> connection is Asynchronous (fast download, slow upload) then it is going
> to run at the slower "upload" speed of the connection minus the protocol
> overhead (mentioned in #1)and minus whatever bandwidth is already being
> used up by other things.
>
> So,...do some speed tests to see what you are dealing with. Anything less
> that about 1,200 mbps is going to perform like crap (IMO).
>
> One of the many speed test sites is http://www.speakeasy.net/speedtest
>
> The next important thing to remember is that there are *two* Internet
> connections involved with VPN,...one at each end,...it only takes one of
> them to screw it up. It is only going to run at the slowest speed of the
> slowest connection, minus the other bandwidth eating factors mentioned
> above.
>
> --
> Phillip Windell
> www.wandtv.com
>
> The views expressed, are my own and not those of my employer, or
> Microsoft,
> or anyone else associated with me, including my cats.
> -----------------------------------------------------
>
>

Hello,

Bandwidth perf: iperf (open source)
Latency: smokeping (open source)

just my 2 cents,

Spin a écrit :
> How would I test the speed between two buildings which are in two
> geographically separate cities in my site-to-site VPN network?
>
> "Phillip Windell" <(E-Mail Removed)> wrote in message
> news:(E-Mail Removed)...
>> "Spin" <(E-Mail Removed)> wrote in message
>> news:(E-Mail Removed)...
>>
>>> Well, consider the case of a dedicated site-to-site VPN, which is an
>>> Internet technology designed to make remote networks "look" local.
>>> In this case, we have a LAN which is really a WAN but made to look
>>> like a LAN. These are the case for which I speak.
>>
>> In that case it is the Internet. The VPN Tunnel is not "real" it is
>> an imaginary thing,...a "logical" thing. So they are just "packets"
>> going over the Internet,..what is ecapsulated within them is not
>> relevant. VPN by nature is slower than regular Internet traffic for
>> two reasons:
>>
>> 1. The required encapsulation and encryption has a cost in performance
>>
>> 2. VPN runs Synchronous (same speed both ways), so if the Internet
>> connection is Asynchronous (fast download, slow upload) then it is
>> going to run at the slower "upload" speed of the connection minus the
>> protocol overhead (mentioned in #1)and minus whatever bandwidth is
>> already being used up by other things.
>>
>> So,...do some speed tests to see what you are dealing with. Anything
>> less that about 1,200 mbps is going to perform like crap (IMO).
>>
>> One of the many speed test sites is http://www.speakeasy.net/speedtest
>>
>> The next important thing to remember is that there are *two* Internet
>> connections involved with VPN,...one at each end,...it only takes one
>> of them to screw it up. It is only going to run at the slowest speed
>> of the slowest connection, minus the other bandwidth eating factors
>> mentioned above.
>>
>> --
>> Phillip Windell
>> www.wandtv.com
>>
>> The views expressed, are my own and not those of my employer, or
>> Microsoft,
>> or anyone else associated with me, including my cats.
>> -----------------------------------------------------
>>
>>
>

"Spin" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> How would I test the speed between two buildings which are in two
> geographically separate cities in my site-to-site VPN network?

Copy a file between the two and time it. Then do the math against the file
size.

Just remember that the file size is in Bytes and the bandwidth speed is in
Bits,...it takes 8 Bits to make a Byte.

So 100 mbps (bits) of line speed = 12.5 MBps (bytes)

So a file that is 25meg in size (bytes) should take two seconds over a 100
mbps line if the whole line's bandwidth is dedicated to the task. But in
"real life" nothing ever max's out the line during a transfer,...there is
*always* some left over bandwidth,...so it would really take more than two
seconds. Remember that if something used all the bandwidth for a single
action then only one thing (user?) could ever use the Line at one
time,...and we know that this is not how it works out in real
life,...multiple things happen over the line at the same time. So in the
end all you are going to get is a general idea of how well it is doing.

Anyone is welcomed to correct my "math" if I screwed it up. I tend to "poke
and eye out" with math if I'm not careful.


--
Phillip Windell
www.wandtv.com

The views expressed, are my own and not those of my employer, or Microsoft,
or anyone else associated with me, including my cats.
-----------------------------------------------------

Also keep in mind that Bandwidth is not the speed.
Thoughput is the speed.

To give a physical illustration of this:

Bandwidth = the number of lanes on a road moving in the same direction

Throughput = the speed limit that the road runs at

So a two lane road can move twice and many cars in the same amount of
time,...but both cars still move at the same speed. So if the road has two
lanes (same direction) with a 60mph limit, it does not mean that if there is
only one car on the road it can run a 120mph.

--
Phillip Windell
www.wandtv.com

The views expressed, are my own and not those of my employer, or Microsoft,
or anyone else associated with me, including my cats.
-----------------------------------------------------