May 2008 - nml-wg - lists.ogf.org

Re: [Nml-wg] Domains
by Freek Dijkstra 31 May '08

31 May '08

Anand Patil wrote: > I agree with Aaron that we should be driven by use cases and not get > bogged down in a theoretical discussion on this issue. Sounds reasonable enough :-). Are there practical use cases for domains in the first deliverable? If not, are there particular use cases for you? If not, here is mine: 1) Global Crossing commisions ships to lay a trans-oceanic fibre. SURFnet leases that fiber for 15 years. SURFnet pays SARA to operate the switch that is connected to that fiber. What is the domain of this fiber? GC? SURFnet or SARA? In TMN terms GC is the legal owner, SURFnet the economic owner, SARA the operator. At the moment I only need/want to model SARA as the OperatorDomain, since they operate the control plane my software likes to talk to. So my use case requires an OperationalDomain, nothing else. Yuri Demchenko, one of my colleagues who is going to present next week, may disagree, since he want to couple policy descriptions to NML, and he may need to model the SURFnet PolicyDomain. Freek

1 0

NML-WG agenda for next week.
by Paola Grosso 29 May '08

29 May '08

Dear all. Most of our time will be devoted to discussion and brainstorming. There have been also a few requests for talks and this is the current agenda: - Monday June 2 @ 3.45pm-5.15pm F. Dijkstra (UvA) "Multi-layer path finding in NDL" (15 minutes) Y. Demchenko (UvA) "Topology descriptions and (multidomain) access control policy" (20 minutes) - Tuesday June 3 @11.00am-12.30pm V. Reijs (HEAnet) "AutoBAHN Stitching Framework" (20 minutes) See you in Barcelona, Paola

1 0

domains and the Stitching Framework (SF)
by Victor Reijs (work) 29 May '08

29 May '08

Hello all of you, I am looking at the types (classes) of domains which might be important when working on Stitching (which is trying to connect different domains together by checking if they are or can be made compatible). Work has been done as part of AutoBAHN (GN2 JRA3 activity). The document that describes this Stitching Framework (SF) is here: http://www.geant2.net/upload/pdf/GN2-07-066v5-DJ3-5-3-Report_on_Testing_of_… This is already an 'old' document (1 year) and was also presented at GLIF in Japan (2006): http://www.glif.is/meetings/2006/controlplane/reijs-jra3.pdf <remember this is all before I knew UML, so it is not using that right UML terminology yet!> I talked with Freek and we discussed the domains that could be important for NML and I proposed to make an overview of the 'domains' (classes and objects in UML language) which are handy (but not always essential!) in the SF. The SF is just one view of reality (a use case), so there might be less classes (more objects) or more classes (less objects) when looking at all use cases of domains. At least the ones I discuss were in some way handy when talking about stitching (the ones with a * at the end look to be important for path finding and stitching). The below list comes from the above GEANT2 document (section 2.2.1) and I tried to UML-ize it (but I am really a novice on UML!). <I am still not 100% confident when to use classes, objects, attributes or values in UML, I am working on that> I tried to check the terminology used within AutoBAHN and IDC (but I did was not able to fully cover this). It also is written from the view of Bandwidth on Demand (circuits), but it can just as well work in packet switched networks, application environment and/or political levels;-) So anywhere compatibility is needed.. Remember I might use IDC and AutoBAHN as examples, but they might not have implemented it!!! I also did not check it with NMWG doc. 'Hierarchy of Characteristics', which is I assume: http://www-didc.lbl.gov/NMWG/docs/draft-ggf-nmwg-hierarchy-02.pdf ? The domain classes that play a some role in the use case, which uses the SF, are : • Termination domain This domain is where the possible path begins or ends. At least two Termination domains must exist in each path. A Termination domain can be a single workstation or a full-blown network (with demand for multiple paths). The Termination domain is not managed by the service (like AutoBAHN/IDC). It has only one 'side' (egress at start of the path and ingress at end of the path). • User domain* [AutoBAHN: UserAccessPoint?] A User domain is a kind of the Termination domain and it is where the actual User’s system/network resides. • Linking domain* [AutoBAHN: External?] A Linking domain has known interfaces/protocols/processes and is one of the domains that make up the possible path between two Termination domains. If a proposed path (LIDP: Loose Inter-Domain Path) is possible all Linking domains together will form the actual path (SIDP: Strict Inter-Domain Path) (like AutoBAHN or IDC). A linking domain has two 'sides' (ingress and egress) • Home domain [AutoBAHN: NA] The Home domain is the domain that will accept the BoD request from a User. The Home domain does not have to be an actual Linking domain in the path; the Home domain is just the domain where the User sends his/her BoD request. So the Home domain acts as a proxy to the service and determines the Source Domain). • Source domain* [AutoBAHN: HomeDomain?] This is the first Linking domain that is part of the possible path. It might starts the investigation if a the BoD request can be made. • Destination domain* [AutoBAHN: LastDomain?] This is the last Linking domain that is part of the possible path. It might do the path finding and feasibility evaluation of the proposed path based on all information gathered from the Linking and Termination domains. • Technology domain Linking domains can use a certain technology (such as pure optical circuits, SDH, SONET, Ethernet, etc.) in its switching/routing core. A Technology domain can allow ports to have a different technology then the Core (so adaptation taking place in the actual interface): e.g. an SDH Technology domain can have Ethernet ports. • Administrative domain Networks fall under the strategic, tactic and operational responsibility of an entity e.g. NREN, operator, etc. Such an entity is called an Administrative domain. (Remember sometimes these three responsibilities can be done by three parties) • Provisioning domain A Provisioning domain is a Linking domain, which is defined by its provisioning methodology instead its technology. E.g. some NRENs (Administrative domain) aggregate their networks into one Provisioning domain. An example is HEAnet with its native Ethernet and an L2 MPLS VLL networks (two different Technology domains), but these two use the same NorthBound interface to form one Provisioning domain (using one tool: BLUEnet tool). Other objects of Provisioning domains are networks that use certain types of control plane or common management tool (aka north bound interface); like GMPLS, UCLP, DRAC, human NOC • Neighbouring domain A Neighbouring domain is a Linking/Termination domain that is physically connected (layer 0) to an adjacent Linking domain for the path. Neighboring domain is a Peering domain at layer 0. The proposed path (LIDP) is build of multiple Neighbouring domains. • Peering domain A Peering domain is the next domain where a certain protocol layer is terminated. This does not have to be the Neighbouring domain. For example two IP workstations interconnected with Ethernet network, the Peering domain of the IP layer are at the two workstations (User domains) which are not the directly connected Neighbouring domain. • Aggregated domain One can groom/group Administrative/Technology/Provisioning domains together over a large community and if that community acts as a single entity then we call this an Aggregated domain. UCLP is e.g. defined over many organisations, so it can be seen as a single Aggregated domain. This can also be used for groomed/grouped together networks for Premium IP (PIP) or (G)MPLS. An Aggregated domain could be a Termination or a Linking domain. Hope this helps in the discussion. I am planning to be by phone at the Tuesday meeting of OGF-NML. All the best, Victor -- Victor Reijs, Network Development Manager HEAnet Limited, Ireland's Education and Research Network 1st Floor, 5 George's Dock, IFSC, Dublin 1 Registered in Ireland, no 275301 tel: +353-1-660 9040 fax: +353-1-660 3666 web: http://www.heanet.ie/

1 0

Reference Model WG
by Freek Dijkstra 23 May '08

23 May '08

Dear chairs (Paola and Martin), Is any of you familiar with the Reference Model WG in the OGF? I only found out about it just when looking through the agenda of OGF 23, and noticed that in the past they also had discussions on for example "AdminDomain" [1], even though their goal seems mostly in describing computing and storage elements, not so much network elements. I have no clue if there is overlap, and how much, but I like to find out. Unfortunately, their meeting collides with our second NML-WG meeting (Tue 11:00-12:30). Two questions: Is there overlap? Should we collaborate (or at least more closely inform each other what we do)? Should we try to reschedule our 2nd meeting? Regards, Freek Dijkstra [1] e.g. http://www.ogf.org/pipermail/rm-wg/2007-July/000081.html

1 0

CFP GridNets 2008
by Pascale VICAT-BLANC 09 May '08

09 May '08

GridNets 2008 ************* Second International Conference on Networks for Grid Applications Beijing, China, October 8-10, 2008 http://www.gridnets.org/ Sponsored by ICST Technically co-sponsored by Create-Net and EU-IST All accepted papers will be published by Springer in the ICST Lecture Notes (LNICST) series. The proceedings will be published by Springer and will be available both as paper-based copies and via Springerlink. The GridNets conference series is an annual international meeting that provides a focused and highly interactive forum where researchers and technologists have the opportunity to present and discuss leading research, developments, and future directions in the Grid networking area. The goal of this event is to serve as both the premier conference presenting best Grid Networking research and a forum where new concepts can be introduced and explored. Previous events in this series were: GridNets 2004 in San Jose (USA), GridNets 2005 in Boston (USA), GridNets 2006 in San Jose (USA) and GridNets 2007 in Lyon (France). All these events have been successful in attracting high quality papers and a wide international participation. From the first event through the fourth, it was known as the GridNets Workshop affiliated with the IEEE BroadNets conference series. From last year on, it has been offered as an independent conference. Grid developers and practitioners are increasingly realising the importance of efficient network support. Entire classes of applications would greatly benefit by a network-aware grid middleware, able to effectively manage the network resource in terms of scheduling, access and use. Conversely, the peculiar requirements of grid applications provide stimulating drivers for new challenging research towards the development of grid-aware networks. Cooperation between grid middleware and network infrastructure driven by a common control plane is a key factor to effectively empower the global grid platform for the execution of network-intensive applications, requiring massive data transfers, very fast and low- latency connections, and stable and guaranteed transmission rates. Large e-science projects, as well as industrial and engineering applications for data analysis, image processing, multimedia, or visualisation just to name a few are awaiting an efficient grid network support. They would be boosted by a global grid platform enabling end-to-end dynamic bandwidth allocation, broadband and low- latency access, interdomain access control, and other network performance monitoring capabilities. Scope * Network architectures and technologies for grids; * Integration of advanced optical networking technologies into the grid environment; * End to end lightpath provisioning software systems and emergent standards; * The network as a first class grid resource: network resource information publication, brokering and co-scheduling with other resources; * Interaction of the network with distributed data management systems; * Network monitoring, traffic characterisation and performance analysis; * Inter-layer interactions: optical layer with higher layer protocols, integration among layers; * Experience with pre-production optical network infrastructures and exchange points; * Peer-to-peer network enhancements applied to the grid; * Network support for wireless and ad hoc grids; * Data replication and multicasting strategies and novel data transport protocols; * Cluster and high performance networks; * Networks for Data Centers; * Networks & Cloud computing issues; * Fault-tolerance, self healing networks; * Security and scalability issues when connecting a large number of sites within a virtual organization VPN; * Simulations; * New concepts and requirements that may fundamentally reshape the network evolution. Keynote Speakers To be defined. Important Dates * Workshop Proposal Deadline: 30 May 2008 * Paper Submission Deadline: 13 June 2008 * Paper Acceptance Notification: 8 August 2008 * Final paper submission: 1 September 2008 * Conference: Beijing during October 8-10, 2008 >> For submission, please refer to http://gridnets.org/authkit.html Organizing Committee -General co-chairs: * Chris Edwards (Lancaster University) * Michael Welzl (University of Innsbruck) * Junsheng Yu (Beijing University of Posts and Telecommunications) -Local co-chairs: * Yuan'an Liu (Beijing University of Posts and Telecommunications) * Tongxu Zhang (China Mobile Group Design Institute Co. Ltd.) * Shaohua Liu (Beijing University of Posts and Telecommunications) -Program Committee co-chairs: * Pascale Vicat-Blanc Primet (INRIA, ENS-Lyon) * Joe Mambretti (Northwestern University) * Tomohiro Kudoh (AIST, Japan)

2 1