FAQ

From WiL

Jump to: navigation, search

Have a question? Answers to frequently asked questions (FAQ) are below.


Contents

[edit] What is WAN-in-Lab?

WAN-in-Lab is a wide area network in a single laboratory. It is an experimental networking testbed aimed at developing, testing and evaluating new communications protocols and technologies.

Unlike emulated testbeds based entirely on dummynet, netem, Emulab or similar technologies, it uses real carrier-grade WAN equipment, and spools of fibre to produce delays. This seeks to reproduce a real production environment more closely.

[edit] What kind of hardware is in WAN-in-Lab?

See:

[edit] What kind of experiments can be run on WAN-in-Lab?

WAN-in-Lab was originally designed for testing flow control protocols, such as FAST. The current hardware allows topologies of up to two hops with variable delays, and multiple 1Gbps bottlenecks.

WAN-in-Lab can also be used to investigate the impact of different scheduling algorithms, or passive or active network measurement algorithms.

WAN-in-Lab is being developed for benchmarking network protocols with the convenience of remote access.

[edit] What is the difference between WAN-in-Lab and a real network?

WAN-in-Lab works in a totally controlled environment, allowing researchers to know exactly what routing and traffic is on the network. By being geographically localised, it also simplifies measurement of aspects of network performance.

[edit] How does WAN-in-Lab achieve long delays with limited fiber?

WAN-in-Lab can provide a delays between routers, up to a total of over 120ms delay at 2.5Gbps, equivalent to trans-Atlantic distances. There are also servers configured as DummyNets to provide longer delays, equivalent to Trans-Pacific or satellite delays. WAN-in-Lab achieves long delays by using loopbacks to cause the signal to pass over each fibre up to 16 times. (See technical FAQ section below for details.)

[edit] How can I get access to WAN-in-Lab?

WAN-in-Lab can be used in either of two ways. Those who simply want to test their own TCP stacks on WAN-in-Lab, you can register for a benchmarking account. This will let them upload your kernel, book time, and evaluate TCP protocols by running standard tests.

For more detailed experiments, users can obtain login accounts, configure the network themselves and their own tests. Access is primarly via ssh to wil-ns.cs.caltech.edu, although some access is available via remote desktops.

For more information, preview Introduction to new users and send mail to wil@caltech.edu

[edit] How can I book time on WAN-in-Lab?

Follow the booking instructions via the website. Bookings should be made in advance, so that others can organise their schedules around your booking. When your time is booked, you can ssh to wil-ns.cs.caltech.edu to use the facilities.

[edit] Who else has used WAN-in-Lab?

WAN-in-Lab has so far been used by researchers in NetLab, and other researchers in the Computer Science department of Caltech.

[edit] I have used WAN-in-Lab for my paper. How should I acknowledge it?

Please add the following acknowledgement: " We acknowledge the use of Caltech's WAN in Lab facility funded by NSF (through grant EIA-0303620), Cisco ARTI, ARO (through grant W911NF-04-1-0095), and Corning."

[edit] I am from a for-profit organization. Can I use WAN-in-Lab?

Yes. Refer to the Guidelines for Commercial Use.

[edit] Where is WAN-in-Lab located?

WAN-in-Lab is housed at the California Institute of Technology, Pasadena.

[edit] Technical questions

[edit] How do I start learning how to use WiL?

Follow the recommendations on the tutorial page and consult people at WiL.

[edit] How does WiL achieve long delays - (technical details)?

Three ONSs each have one four-wavelength OC192 WDM card and four OC48 to OC192 TDM MUX cards, each with an OC48 SFP. Three have amplifiers. Each router is connected to one OC48 port in an SFP, and the other ports are all looped back from one TDM/WDM slot to another. Light comes from the router, gets multiplexed, goes through a fibre spool, gets amplified nd returns through another fibre spool and then gets demultiplexed. After demultiplexing, it is either remultiplexed, or sent to one of the routers. The exact topology depends on the setting of the switch, which allows us to vary the delay by switching segments in or out. (See FAQ section for overview description.)

[edit] Why can't I connect to Servers A1 to C2?

Some servers are automatically turned off when there is no current project. Before turning a server on, go to <http://wil-ns.cs.caltech.edu/book> to book time. (Otherwise, it will be automatically turned off again.)

Once your project is running, you can turn the server on using the wil.reboot-force script:

wil.reboot-force  server<x>

If the WiL-FS server is down, where the WAN-in-Lab website is stored, follow the rebooting instructions on the WiL-FS page or contact us.

[edit] Why can't I log in to servers A1 to C2?

They're turned off when no project is booked.

[edit] How can I use SNMP?

SNMP allows router settings to be queried or set.

[edit] I've just installed a kernel. Why won't my server boot?

See Troubleshooting#New server doesn't boot.

[edit] See also

Views
Personal tools