| 1 | [[PageOutline]] |
| 2 | |
| 3 | = GENI Shadownet Project Status Report = |
| 4 | |
| 5 | Period: Post GEC 17 Report |
| 6 | |
| 7 | == I. Major accomplishments == |
| 8 | |
| 9 | The following highlights our accomplishments |
| 10 | during the last reporting period. |
| 11 | |
| 12 | === A. Milestones achieved === |
| 13 | |
| 14 | All required milestones have been completed. At this point our efforts have |
| 15 | focused on enhancements and improved interoperability with other tools and |
| 16 | services. Two major accomplishements this reporting period include: |
| 17 | |
| 18 | * support for ExoGENI Racks, and |
| 19 | * integration with the GENI Portal. |
| 20 | |
| 21 | === B. Deliverables made === |
| 22 | |
| 23 | * Continued to modify and enhance our instrumentation and measurement code and the GENI Desktop. The updated code is regularly contributed to the GEMINI repositories. |
| 24 | |
| 25 | == II. Description of work performed during last quarter == |
| 26 | |
| 27 | The following provides a description of the progress made during the last reporting period. |
| 28 | |
| 29 | |
| 30 | === A. Activities and findings === |
| 31 | |
| 32 | During this reporting period we made significant progress in three primary |
| 33 | areas; namely GENI Rack support, GENI Portal integration, and GENI Desktop |
| 34 | enhancements. |
| 35 | |
| 36 | First, we continued to work with the groups at Utah and the GPO to further |
| 37 | integrate our InstaGENI rack into GENI and the GENI Openflow network. As |
| 38 | part of this work, we performed extensive testing of the rack, using it |
| 39 | almost exclusively in all of our experimentation with the GENI desktop and |
| 40 | GEMINI. In addition to working with our InstaGENI rack (and the other |
| 41 | InstaGENI racks that have been coming online), we also enhanced our |
| 42 | instrumentation and measurement code to work with ExoGENI racks. Because |
| 43 | there are significant and fundamental differences between InstaGENI and |
| 44 | ExoGENI, porting our code to ExoGENI was a challenge and involved close |
| 45 | cooperation with the ExoGENI team. Different RSPEC formats/extensions, |
| 46 | the node OS used, software preinstalled on nodes, internal and external |
| 47 | IP address assignments (or the lack thereof), the ways |
| 48 | for the code on an experimental node to discover its context, and other differences made |
| 49 | porting to the ExoGENI environment non-trivial. |
| 50 | Despite these challenges, |
| 51 | we now have an intial version working that demonstrates basic interoperability with |
| 52 | ExoGENI racks. There are still several issues to be addressed, but users of ExoGENI slices |
| 53 | should now be able to leverage the GENI Desktop/GEMINI tools and interfaces. |
| 54 | |
| 55 | Second, we worked with the GENI portal team to provide interoperability |
| 56 | between the GENI Portal and the GENI Desktop. A key goal here was to provide |
| 57 | autologin of portal users into the GENI Desktop. The key issue, which |
| 58 | is a well-known issue, is the need for |
| 59 | tools to "speak as" a user, rather than "speak for" a user. Consequently, |
| 60 | the GENI Desktop needs to use the user's credentials to make GENI AM API |
| 61 | calls and act on the user's behalf. To address this issue, we worked with |
| 62 | the Portal team to develop a protocol between that would pass the needed |
| 63 | information from the GENI Portal to the GENI Desktop. In the future, the |
| 64 | plan is to pass "speaks for" credentials. It should be noted, that this |
| 65 | exchange of information was also combined with the ability to seamlessly move back |
| 66 | and forth between the GENI |
| 67 | Portal web pages to GENI Desktop web pages. We also changed |
| 68 | the GENI Desktop login page to redirect to the GENI Portal login page so that |
| 69 | GENI Portal account information could be used to login to the GENI Desktop. |
| 70 | After working through many challenges with the Portal team and |
| 71 | making adjustments to both the GENI Portal and GENI Desktop, users are now |
| 72 | able to login to the GENI Portal and then move freely between the two tools. |
| 73 | |
| 74 | The interoperability between the GENI Desktop and the GENI Portal was |
| 75 | complicated by the fact that there is currently no standard for slice |
| 76 | authorities. |
| 77 | The lack of a common interface to slice authorities makes it difficult for |
| 78 | tools like the GENI Desktop to query the slice authority to find information |
| 79 | about a slice -- information needed to list slices and provide a graphical |
| 80 | image of the topology. At present, the GENI Desktop is able to retrieve this |
| 81 | infomation using a ProtoGENI-like SA API, but this does not leverage all the |
| 82 | features of the GENI Portal SA, nor will it work with SAs that do not support |
| 83 | these APIs. A common SA API would clearly help to facilitate better |
| 84 | interoperability. |
| 85 | |
| 86 | Third, we continued to enhance the GENI Desktop, making improvements to its |
| 87 | look-and-feel, functionality, and extensibility. We also refactored its code |
| 88 | to make it more robust to failures, and to make it easier to write new |
| 89 | modules for the Desktop. In particular, we developed an initial prototype of |
| 90 | a GENI Desktop module that hopefully could be used by experimenters in the |
| 91 | future to write their own extensions to the GENI Desktop. As part of the |
| 92 | change to the look-and-feel, we completely redesigned the entry page which |
| 93 | lists the slices. The goal of our redesign was to be able to extend the GENI |
| 94 | Desktop abstraction of "select and then apply an operation" to slices as |
| 95 | well. Users can now select one or more slices and then apply an |
| 96 | operation to those slices such as extend the lifetime of the slice(s), |
| 97 | instrumentize the slice(s), get the status of the slice(s), or delete the |
| 98 | slice(s). We have already found these features to be very helpful in our own |
| 99 | experimentation and testing of the GENI Desktop. |
| 100 | |
| 101 | We demonstrated the new GENI Desktop system and its integration with |
| 102 | ExoGENI and the GENI Portal at GEC 17. We also gave a presentation on the |
| 103 | GENI Desktop to students at the Second GENI Research and Educational |
| 104 | Experiment Summer Camp (GREE-SC 2013). In both conferences/workshops we gave |
| 105 | a hands-on tutorial that first introduced the concepts and abstractions |
| 106 | underlying the GENI Desktop, and then presented the attendees with a set of |
| 107 | tasks to complete using the GENI Desktop to help strengthen their |
| 108 | understanding and ability to utilize the various features of the GENI |
| 109 | Desktop. |
| 110 | |
| 111 | As in the past, we continued to manage and operate the |
| 112 | Juniper routers that comprise the Shadownet aggregate, making these resources |
| 113 | available to users for use in their experiments. |
| 114 | |
| 115 | === B. Project participants === |
| 116 | |
| 117 | The following individuals are involved with the project in one way or another: |
| 118 | * Jim Griffioen - Project PI (Kentucky) |
| 119 | * Zongming Fei - Project Co-PI (Kentucky) |
| 120 | * Kobus van der Merwe - Project Co-PI (was AT\&T, now at Utah) |
| 121 | * Eric Boyd - Subcontract Lead (Internet2) |
| 122 | * Brian Cashman - Network Planning Manager (Internet2) |
| 123 | * Lowell Pike - Network administrator (Kentucky) |
| 124 | * Hussamuddin Nasir - Technician/Programmer (Kentucky) |
| 125 | * Charles Carpenter - Researcher/Programmer (Kentucky) |
| 126 | * Jeremy Reed - Research Assistant (Kentucky) |
| 127 | * Emmanouil Mavrogiorgis - Research Staff (AT\&T) |
| 128 | |
| 129 | === C. Publications (individual and organizational) === |
| 130 | |
| 131 | The following publication describes a network hypervisor service that was an |
| 132 | outgrowth of our work on GENI and automated slice setup. |
| 133 | |
| 134 | * Network Hypervisors: Managing the Emerging SDN Chaos, Shufeng Huang, James Griffioen, International Conference on Computer Communications and Networks (ICCCN 2013) Track on Network Architectures and Clean-Slate Designs (NACSD), July 30 - August 2, 2013 |
| 135 | |
| 136 | |
| 137 | === D. Outreach activities === |
| 138 | |
| 139 | * Jim Griffioen and Hussamuddin Nasir (together with colleagues from Indiana and Utah) gave a demonstration and tutorial at GEC 17 about InstaGENI, the GENI Desktop and the GEMINI I\&M system. |
| 140 | * We also presented during the demo session at GEC 17 showing the latest version of the GENI Desktop and its features. |
| 141 | |
| 142 | === E. Collaborations === |
| 143 | |
| 144 | Most of our collaborations have been with the Shadownet team. It involves |
| 145 | participants from Kentucky, Utah (previously at AT\&T), and Internet2, but we |
| 146 | have also collaborated closely with our InstaGENI and perfSONAR/LAMP/GEMINI |
| 147 | colleagues. |
| 148 | |
| 149 | === F. Other Contributions === |
| 150 | |