Alrighty. For anyone interested and those that have been following the Hub/switch thread, the following are the tests we are looking to do:
As much as possible, the testing will be performed using a typical consumer-grade eight-port switch, and a typical consumer-grade eight-port hub. Both will be recent models of popular brands. There's a minor problem, and, if y'all will accept the work-around, we'll do it. Noone I know has a consumer-grade 100Meg HUB. We have a couple commercial-grade 100MEG-only hubs in the Lab...so if you want to see 100meg hub against 100meg switch, the 100Meg hub will be either a 3COM or Intel commercial-grade hub. Otherwise, to keep things "apples-to-apples," the testing will have to be in the 10Meg relm.
Think about it and let me know by Friday (the proposed test day, still looking good).
The test (Smartbits):
One stream in, one stream out
Two streams in, two streams out
Four streams in, four streams out
Two streams in, one stream out
Four streams in, one stream out
Seven streams in, one stream out (<--wanna do this?)
The basic test suite from NetIQ has testing for throughput, latency, and packet drop. Throughput and latency will run for 10 seconds per iteration, using five packet sizes (64, 256, 512, 1024, and 1500 bytes). The packet-drop test will run for two minutes, using the same packets sizes.
Chariot
Four WINTEL 866MHz PCs plus one server will be attached to the device-under-test (DUT). Each PC will be an endpoint, the server will be an endpoint. Each PC will have one or more emulated streams simulating a large file transfer (and possibly a web session, email, or streaming video) to the server. In addition, each pair of PC Endpoints will run a concurrent stream emulating a print session, file transfer, or game stream. The number of actual sessions and the type of emulated stream will be adjusted to induce a moderate load on the DUT (or, you tell me, wanna crush it? Take it light?).
Before the test, all jumpers are scanned with a WaveTek cable scanner to ensure Category 5 compliance. The PCs will all have 10/100 NICs (Intel Pro 100s), the server will also have an Intel Pro100 NIC. Each PC has 512Meg of RAM. The Dual-Xeon 933 server has 1 GB of RAM. The Endpoint software runs as a service in the background on all machines.
My intention is to publish a BRIEF summary of the results in this forum, with the complete and unedited output from the Smartbits test and Chariot test available from a web page and/or FTP server.
The plan (so far) is to run the tests Friday afternoon. I'm doing this at work, so if any of that nasty work-stuff happens, I may have to push it back to next week.
In the meantime, if I've omitted something from this plan, or if there's something you really want to see, post it in this thread and we'll try to get it into the test.
Let me know .....
Scott
As much as possible, the testing will be performed using a typical consumer-grade eight-port switch, and a typical consumer-grade eight-port hub. Both will be recent models of popular brands. There's a minor problem, and, if y'all will accept the work-around, we'll do it. Noone I know has a consumer-grade 100Meg HUB. We have a couple commercial-grade 100MEG-only hubs in the Lab...so if you want to see 100meg hub against 100meg switch, the 100Meg hub will be either a 3COM or Intel commercial-grade hub. Otherwise, to keep things "apples-to-apples," the testing will have to be in the 10Meg relm.
Think about it and let me know by Friday (the proposed test day, still looking good).
The test (Smartbits):
One stream in, one stream out
Two streams in, two streams out
Four streams in, four streams out
Two streams in, one stream out
Four streams in, one stream out
Seven streams in, one stream out (<--wanna do this?)
The basic test suite from NetIQ has testing for throughput, latency, and packet drop. Throughput and latency will run for 10 seconds per iteration, using five packet sizes (64, 256, 512, 1024, and 1500 bytes). The packet-drop test will run for two minutes, using the same packets sizes.
Chariot
Four WINTEL 866MHz PCs plus one server will be attached to the device-under-test (DUT). Each PC will be an endpoint, the server will be an endpoint. Each PC will have one or more emulated streams simulating a large file transfer (and possibly a web session, email, or streaming video) to the server. In addition, each pair of PC Endpoints will run a concurrent stream emulating a print session, file transfer, or game stream. The number of actual sessions and the type of emulated stream will be adjusted to induce a moderate load on the DUT (or, you tell me, wanna crush it? Take it light?).
Before the test, all jumpers are scanned with a WaveTek cable scanner to ensure Category 5 compliance. The PCs will all have 10/100 NICs (Intel Pro 100s), the server will also have an Intel Pro100 NIC. Each PC has 512Meg of RAM. The Dual-Xeon 933 server has 1 GB of RAM. The Endpoint software runs as a service in the background on all machines.
My intention is to publish a BRIEF summary of the results in this forum, with the complete and unedited output from the Smartbits test and Chariot test available from a web page and/or FTP server.
The plan (so far) is to run the tests Friday afternoon. I'm doing this at work, so if any of that nasty work-stuff happens, I may have to push it back to next week.
In the meantime, if I've omitted something from this plan, or if there's something you really want to see, post it in this thread and we'll try to get it into the test.
Let me know .....
Scott
