[K12OSN] Testing actual network speeds

[Bob McCaa]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Terrell Prudé Jr. wrote:
> Dan Young wrote:
>> Carl Keil wrote:
>>   
>>> Can anyone suggest a way to test my LAN speeds that I'm using for
>>> K12LTSP?  Screen redraws are jerky and slow on some of my clients.  I've
>>> always suspected my network cabling wasn't quite right, but I don't know
>>> how to test it.
>>>     
>>
>> You can check for errors and find out what speed your NICs are
>> negotiating with ifconfig and ethtool. Substitute your ethernet
>> interfaces for eth0 (i.e. it may be eth1, etc.)
>>
>> "ifconfig eth0 | grep errors"
>> Any number of errors is not good, especially if that number rises much
>> over time.
>>
>> "ethtool eth0"
>> Look at speed and duplex.
>>
>> Throughput can be monitored with iptraf (install w/ yum). The detailed
>> interface statistics can tell you incoming/outgoing kbps.
>>
>> If your switches are "managed", you might want to use their command-line
>> or web interfaces to look for CRC errors or similar. With these kind of
>> problems, it pays to be methodical and check every device/port between
>> the server and the client.
>>
>>   
> 
> And to actually generate some test traffic, I take a reasonably powerful
> box (dual-Athlon desktop, Core Duo laptop, or whatever) and use the
> following command:
> 
> [microman at takhisis ~]$ su root
> Password:
> [root at takhisis microman]# ping -f -s 15000 w.x.y.z
> 
> where w.x.y.z is some other pretty beefy box on the other side of the
> link.  What you're doing here is called a "flood" ping, and you're
> specifying a packet size of 15,000 bytes/packet.  I will do this until I
> fill up the link.  I've actually pegged an OC-12c with multiple such
> ping sessions.  I would have, say, four of these from Box A pinging Box
> B, and I'd have another four or so from Box B going to Box A.  So, you
> should have no trouble filling up a 100Mbps client link.
> 
> Here's another way:  fire up full-screen TuxType or TuxMath sessions on
> all of your clients, simultaneously.  This *WILL* generate a boatload of
> traffic.  All you need are 15 simultaneous sessions of this to peg the
> server's Gig-E link (one of them will use 73Mb/sec).
> 
> Then, as Dan mentions above, check out the port statistics on your
> switch, assuming that it's a managed switch (e. g. BayStack 450T, Cisco
> Catalyst, or Amer.com SR48G2i).  If you're not using a managed switch,
> it actually would be a wise investment; 24-port BayStack 450T's with a
> fiber Gig-E interface can be had on eBay for about $200.  They work
> really well.
> 
> --TP
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> K12OSN mailing list
> K12OSN at redhat.com
> https://www.redhat.com/mailman/listinfo/k12osn
> For more info see <http://www.k12os.org>

Iperf is a good tool for point to point bandwidth testing.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFFnVgz3zXhtRrNjggRApXUAJ4+xtpmQIUNJF8AeXvKHwsDoXEMRACghoQ+
h7nXspPWBLUFchHqygvt6Yk=
=a9zw
-----END PGP SIGNATURE-----