cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
2097
Views
0
Helpful
15
Replies

Measuring Actual Bandwidth / DS3 Lines

Jkloza_2
Level 1
Level 1

Hello all,


I'm currently in the process of doing a bandwidth analysis on 2 WAN links at my current HQ / remote sites.  The topology is pretty straightforward, two 3845 routers at HQ, and two 3845 routers at my remote site, both with P2P DS3 interfaces between them.  From each router we then connect directly into a 6509 at the HQ, and a 4506 at my remote site where we are using equal cost load sharing (EIGRP) to split the load / give us redundancy between the sites.

I'm just looking for a good test for how well the circuits are performing.  I've done some basic file transfer tests, but they're only showing about 4MBPS, which seems quite slow.  I also know that this isn't a true measurement because there are amny different factors involved, like hard drive write speeds, network congestion, etc, that may influence the test.

Anyone out there have some suggestions on a good tool / bandwidth analysis tool?

Thanks in advance for any help!

15 Replies 15

ciscoben2009
Level 1
Level 1

Hello

I normaly copy a ISO file and use PRTG is runs of snmp and set the refresh to a second but you need the paid version

hope that helps

thanks

Ben

sleepyshark
Level 1
Level 1

Grab any SNMP monitoring utility and pull SNMP stats off of your routers...

MRTG is a good one (a little complex to setup on windows if you're not familiar)

Whatsup Golds (trial available)

Solarwinds (trial available)

you'll need to enable SNMP on the routers/switches to allow monitoring traffic.

Thanks,

Sean Brown

http://www.sleepyshark.com

The products above do not measure bandwidth. They display traffic amount.

Banditdh is measure with tools like iperf, and many others.

Yeah, actually they do measure bandwidth.... In fact let's split some hairs here... bandwidth is traffic...

Traffic received/send divided by the Bandwidth available = Utilization....  Anything that logs traffic can calculate bandwidth used/available/utilized.

I'm not splitting hairs, and you are missing the point.

OP want to measure capacity of a recently delivered circuit.

That is done by

1) offering traffic to the cricuit,

2) measuring how much is received within an unit of time.

Common network management tools, do (2) but not 1), so they don't do the job.

vmiller
Level 7
Level 7

Its unlikeley you have < rated bandwidth on the links. you may have a throughput issue.

start poking around your traffic flows, take a look at some workstation configs to see if the window size can

be increased.

I agree with that assessment, I'm sure we have the 44.XX per DS3, so it does seem to be a throughput problem.  We are a Vista shop (unfortunatley), until we migrate to Windows 7 in the coming months.  I've tested disabling of the TCP auto-tuning which sets the MSS at 65535, but the fastest transfer i'm seeing is right around 5MBPS, not 20 or 30 like i'm expecting. 


Any suggestions on tuning the systems?

Before blaming the circuit, measure the UDP performances using a proper tool.

And keep an eye on "show controllers" and "show interface", they have to 100% error free.

I'm not a workstation expert. There does appear to be a wealth of documents and suggestions available on the internet that

address workstation TCP IP stacks.

Paolos point is spot on. Its easy to do a quick check of the controllers, and interface counters, and either:

  a. start there, if you notice CRC's & drops.

  b. move on, since there are no reported problems.

If you are only having issues on one app, focus on that. If the symptoms are more global, gain an understanding of how

the client server apps communicate. Lots of apps tend to be real chatty, which manifests itself as slow throughput.

It can be a full time job.

Understood, I've run through all 4 routers already, 0 CRC errors on my serial interfaces, not problems with the controllers etc.


I'm in the process of getting some netmon tools, but I'm not sure they're going to show me exactly what i want to see.


Thanks - 

Disclaimer

The  Author of this posting offers the information contained within this  posting without consideration and with the reader's understanding that  there's no implied or expressed suitability or fitness for any purpose.  Information provided is for informational purposes only and should not  be construed as rendering professional advice of any kind. Usage of this  posting's information is solely at reader's own risk.

Liability Disclaimer

In  no event shall Author be liable for any damages whatsoever (including,  without limitation, damages for loss of use, data or profit) arising out  of the use or inability to use the posting's information even if Author  has been advised of the possibility of such damage.

Posting

For quick performance testing I use a freebie pcattcp.  It can easily saturate a DS3 link with it's UDP option and can also do TTCP testing.  With this tool, you might first try sending 50 Mbps down a link and see what rate comes out the other side.  Then you might try TTCP.

PS:

If you're doing packet-by-packet equal cost load balancing, you might be causing TCP performance issues.

Joseph,


Currently, I'm running equal cost load sharing (per destination, not per packet).  It does look like there's a TCP throughput issue.  From the quick tests I ran this morning using TTCP, this is what I'm seeing TCP vs UDP:

C:\TCP>pcattcp -t X.X.X.X

PCAUSA Test TCP Utility V2.01.01.14 (IPv4/IPv6)
  IP Version  : IPv4
Started TCP Transmit Test 0...
TCP Transmit Test
  Transmit    : TCPv4 0.0.0.0 -> X.X.X.X:5001
  Buffer Size : 8192; Alignment: 16384/0
  TCP_NODELAY : DISABLED (0)
  Connect     : Connected to X.X.X.X:5001
  Send Mode   : Send Pattern; Number of Buffers: 2048
  Statistics  : TCPv4 0.0.0.0 -> X.X.X.X:5001
16777216 bytes in 17.423 real seconds = 940.38 KB/sec +++
numCalls: 2048; msec/call: 8.711; calls/sec: 117.548

C:\TCP>pcattcp -t -u X.X.X.X

PCAUSA Test TCP Utility V2.01.01.14 (IPv4/IPv6)
  IP Version  : IPv4
Started UDP Transmit Test 0...
UDP Transmit Test
  Transmit    : UDPv4 0.0.0.0 -> X.X.X.X:5001
  Buffer Size : 8192; Alignment: 16384/0
  Send Mode   : Send Pattern; Number of Buffers: 2048
  Statistics  : UDPv4 0.0.0.0 -> X.X.X.X:5001
16777216 bytes in 1.386 real seconds = 11819.88 KB/sec +++
numCalls: 2050; msec/call: 0.692; calls/sec: 1478.928

Thanks -

Disclaimer

The   Author of this posting offers the information contained within this   posting without consideration and with the reader's understanding that   there's no implied or expressed suitability or fitness for any purpose.   Information provided is for informational purposes only and should not   be construed as rendering professional advice of any kind. Usage of  this  posting's information is solely at reader's own risk.

Liability Disclaimer

In   no event shall Author be liable for any damages whatsoever (including,   without limitation, damages for loss of use, data or profit) arising  out  of the use or inability to use the posting's information even if  Author  has been advised of the possibility of such damage.

Posting

Drops (anywhere end-to-end) against TCP flows will very much slow their performance.  Don't recall whether this particular TTCP program provides additional TCP stats, such as retransmits.  (I recall Cisco's IOS hidden TTCP command [only in some/few IOS images] can.)

For TCP to saturate a link, receiving host's RWIN must be at least large enough for the path's BDP (usually not an issue for Windows' clients with the NextGen stack; Vista and later, but often worth checking).

Devices along the way also need to be able to buffer slow start TCP transmission window expansion.

TCP does take time to ramp up to full speed; very dependent on end-to-end latency.  A small enough transmission window might not reach full transmission speed as it could still be ramping up.  Quick test of that, with TTCP, increase total amount of transmission data (for example, increase from 16 MB to 32 MB and see if overall speed jumps; take special note if you monitor transmission rate whether it continues to increase over time of the transmission test).

Maybe run the test for longer time.

.

UDP 11819 KB/s = 94 Mb/s

Doesn't make sense for a DS3.

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Review Cisco Networking products for a $25 gift card