Changes between Version 6 and Version 7 of GENI-in-a-Box
- Timestamp:
- 09/07/12 17:00:54 (12 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
GENI-in-a-Box
v6 v7 5 5 6 6 The figure below is a schematic of the GENI-in-a-Box. The GENI-in-a-Box !VirtualBox VM includes the [http://trac.gpolab.bbn.com/gcf/wiki/Omni Omni] experiment control tool, a clearinghouse used to create slices and an aggregate manager that makes resources available for experimentation. The resources available from the GENI-in-a-Box aggregate manager are up to six virtual machines running Linux. Each of these virtual machines has three NICs. They can be connected to form any topology that does not require more than six nodes and more than three NICs on a node. GENI standard rspecs are used to specify the machines, links and topology required by the experiment. 7 8 All the virtual machines created for you run the same Linux distribution and version. You pick the distribution you want to work with by downloading the appropriate variety of GENI-in-a-Box. At this time there are two flavors of GENI-in-a-Box: A Fedora 15 flavor and a Ubuntu 10.0.4 flavor. 7 9 8 10 You will have root access to the virtual machines used in your experiment. You can use ssh to connect to and log into these machines. … … 18 20 19 21 === Can I move my experiments from GENI-in-a-Box to real resources in the GENI infrastructures? === 20 GENI-in-a-Box resources share a number of similarities with resources from the ProtoGENI aggregate. It will therefore be easiest to move your experiment from GENI-in-a-Box to ProtoGENI. See this page for things to watch for when you move your experiment from GENI-in-a-Box to ProtoGENI. 21 22 GENI-in-a-Box resources share a number of similarities with resources from ProtoGENI aggregates. It will therefore be easiest to move your experiment from GENI-in-a-Box to a ProtoGENI-based aggregate. See this page for things to watch for when you move your experiment from GENI-in-a-Box to ProtoGENI resources. 22 23 23 24 24 25 === What are the limitations of GENI-in-a-Box? === 25 - network fidelity 26 - performance isolation 27 - size 26 Because all of GENI-in-a-Box exists within one !VirtualBox VM it is subject to limitations such as: 27 * Limited performance isolation. The experimenter virtual machines compete with one another and with other processes running on your laptop for CPU and network resources. Software execution times may therefore not be repeatable. 28 * Lack of control over link bandwidth. Some GENI aggregates such as ProtoGENI allow experimenters to specify the bandwidth of the links used in the experiment. GENI-in-a-Box ignores any bandwidth specifications in the request rspec. Additionally, all links in GENI-in-a-Box are bi-directional. 29 * Relatively small numbers of resources available for experimentation. Experiments on GENI-in-a-Box are limited to a relatively small number of nodes (currently six). 30 * Limited of OS options. All experimenter virtual machines inside and instance of GENI-in-a-Box run the same OS. The OS version is determined by the flavor of GENI-in-a-Box you are using. The GENI-in-a-Box aggregate manager ignores any OS specifications in the request rspec. 31 28 32 29 33 === Where can I learn more about the implementation of GENI-in-a-Box? === 30 34 If you are a software developer interested in learning about how GENI-in-a-Box is implemented or interested in enhancing GENI-in-a-Box you'll want to check out ... 31 35 32 For more information on GENI-in-a-Box, email [mailto:vthomas@bbn.com Vic Thomas] 36 === Where can I get help with GENI-in-a-Box? === 37 38 Please email your questions to [mailto:help@geni.net].
