Changes between Version 1 and Version 2 of NetworkRspecMiniWorkshopNotes


Ignore:
Timestamp:
06/26/09 10:40:57 (10 years ago)
Author:
Christopher Small
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • NetworkRspecMiniWorkshopNotes

    v1 v2  
    1 See http://groups.geni.net/geni/wiki/NetworkRSpecMiniWorkshop for sides
    2 
    3 9:15 intro slides (Aaron)
    4 
    5 High level motivation goals for spiral 1. See Aaron's slides. Demonstrate
     1'''See http://groups.geni.net/geni/wiki/NetworkRSpecMiniWorkshop for sides'''
     2
     3== Intro ''Aaron Falk'' ==
     4
     5High level motivation goals for spiral 1. See Aaron Falk's slides. Demonstrate
    66end-to-end slices across representative samples of the major substrates and
    77technologies envisioned in GENI. Goal for each cluster is to demonstrate
     
    1111at the end of spiral 1
    1212
    13 (John Turner) What is an "end"?
     13''John Turner'' What is an "end"?
    1414
    1515Loosely defined, but think from perspective of experimenter.
    1616
    17 (Larry Peterson) Key bar is two or more aggregates sharing a packet?
     17''Larry Peterson'' Key bar is two or more aggregates sharing a packet?
    1818
    1919No, aggregates, not in terms of apckets
    2020
    21 (James Sterbenz) Not single slice across ''all'' of them in year 1, pairwise is
    22 sufficient
     21''James Sterbenz'' Not single slice across ''all'' of them in year 1,
     22pairwise is sufficient
    2323
    2424Only pairwise would be a dissapointment. Really want to show that it's
     
    3737spiral 1?
    3838
    39 (John Turner) When you talk about an end-to-end slice, do you have a prescription for
    40 what the data path looks like?
    41 
    42 You're in cluster B, a candidate slice would include internet2 vlans
     39''John Turner'' When you talk about an end-to-end slice, do you have a
     40prescription for what the data path looks like?
     41
     42You're in cluster B, a candidate slice would include Internet2 VLANs
    4343connecting spp nodes and !GpENI and Stanford.
    4444
    45 (John Turner) Sounds like internet2 vlans on an spp somehow ending up at stanford
    46 
    47 GENI will always have a hodge-podge of connectivity. For now we'll have a set
    48 of preconfigured fixed vlans in internet 2.
    49 
    50 (John Turner) VLANS in I2 are being provided by Rob Ricci, not I2. I don't
    51 want to pick on Stanford per se, but it's not clear how we're going to make
    52 any connection happen.
    53 
    54 (James Sterbenz) Also what sort of experiments will tie all of these things
    55 together, given the disparate technologies
    56 
    57 (Larry Peterson) this is what will come out of discussions today
    58 
    59 (4) is raising two questions -- what capabilities will we have from backbones
    60 in spiral 1
    61 
    62 (Camilo Viecco) are any clusters also using NLR?
    63 
    64 (Guido) there is potentially more than one I2 backbone
    65 
    66 (Rob Ricci) the ones we're handing out today are between individual sites. I2
    67 is providing a 10Gb wavelength and we're putting ethernet switches on top of
    68 that wave. We'll be making VLANs on that wave without any I2 involvement.
    69 
    70 (Ivan Seskar) also an issue when particular locations will be connected to the
    71 wave
     45''John Turner'' Sounds like Internet2 VLANs on an spp somehow ending
     46up at stanford
     47
     48GENI will always have a hodge-podge of connectivity. For now we'll
     49have a set of preconfigured fixed VLANs in internet 2.
     50
     51''John Turner'' VLANS in Internet2 are being provided by Rob Ricci,
     52not Internet2. I don't want to pick on Stanford per se, but it's not
     53clear how we're going to make any connection happen.
     54
     55''James Sterbenz'' Also what sort of experiments will tie all of these
     56things together, given the disparate technologies
     57
     58''Larry Peterson'' this is what will come out of discussions today
     59
     60What capabilities will we have from backbones in spiral 1?
     61
     62''Camilo Viecco'' are any clusters also using NLR?
     63
     64''Guido Appenzeller'' there is potentially more than one Internet2 backbone
     65
     66''Rob Ricci'' the ones we're handing out today are between individual
     67sites. Internet2 is providing a 10Gb wavelength and we're putting
     68ethernet switches on top of that wave. We'll be making VLANs on that
     69wave without any Internet2 involvement.
     70
     71''Ivan Seskar'' also an issue when particular locations will be
     72connected to the wave
    7273
    7374There is a ''general'' problem here. Let's not get lost in the weeds.
    7475
    75 (Larry Peterson) Simple observation. As diverse as these technologies are, we
    76 have IP addresses for all of them. We can fall back to using IP (tunnels) for
    77 everything.
    78 
    79 End users can access GENI experimentation this way. But we have set as a goal
    80 non-IP, not layered over IP connectivity for GENI.
    81 
    82 (Rob Ricci) Maybe we should do with !!GpENI -- get a fiber from the local I2 POP
    83 to our campus.
    84 
    85 (John Turner) From my experience the I2 folks will push back not make this easy
    86 
    87 Yes, we've got some things to work out here. However, to be concrete, those
    88 who have direct connectivity in sprial 1 will be expected to demonstrate the
    89 ability to stitch together VLANs.
    90 
    91 (Ivan Seskar) Can GPO be more involved with these discussions with I2 / NLR?
     76''Larry Peterson'' Simple observation. As diverse as these
     77technologies are, we have IP addresses for all of them. We can fall
     78back to using IP ''tunnels) for everything.
     79
     80End users can access GENI experimentation this way. But we have set as
     81a goal non-IP, not layered over IP connectivity for GENI.
     82
     83''Rob Ricci'' Maybe we should do with !!GpENI -- get a fiber from the
     84local Internet2 POP to our campus.
     85
     86''John Turner'' From my experience the Internet2 folks will push back
     87not make this easy
     88
     89Yes, we've got some things to work out here. However, to be concrete,
     90those who have direct connectivity in sprial 1 will be expected to
     91demonstrate the ability to stitch together VLANs.
     92
     93''Ivan Seskar'' Can GPO be more involved with these discussions with
     94Internet2 / NLR?
    9295
    9396Yes, we have full time staff who would be happy to help work things out with
    94 I2, NLR, etc.
    95 
    96 (Ilya) can we organize and get some shared stimulus money to wire up campuses?
    97 
    98 (Chip) My understanding is that every campus is supposed to make a single
     97Internet2, NLR, etc.
     98
     99''Ilya Baldine'' can we organize and get some shared stimulus money to wire up
     100campuses?
     101
     102''Chip Elliott'' My understanding is that every campus is supposed to make a single
    99103proposal to NSF.
    100104
    101 9:35 Network Cofiguration use case slides (Aaron)
    102 
    103 Sliver creation. First makes reservations of stuff around the edge, but now
    104 needs to interconnect aggregates. (Assumption is that physical connectivity
    105 between these aggregates exists.) Then researcher passes rspec requesting
    106 VLANs between aggregates, then asks for the topology to be set up.
     105== Network Cofiguration Use Case ''Aaron Falk'' ==
     106
     107Sliver creation. First makes reservations of stuff around the edge,
     108but now needs to interconnect aggregates. (Assumption is that
     109physical connectivity between these aggregates exists.) Then
     110researcher passes rspec requesting VLANs between aggregates, then asks
     111for the topology to be set up.
    107112
    108113- do we need a standard method to describe these network coordinates, or are
    109   they just blobs?
     114they just blobs?
    110115
    111116- does it go into the rspec?
    112117
    113118- are there now constraints on the order in which networks can be added to a
    114   slice?
     119slice?
    115120
    116121- how does it work with multiple networks in a series?
     
    123128this morning.
    124129
    125 (John Turner) let's put this in as concrete terms as possible, I have a very difficult
    126 time connecting your abstract diagrams with my cluster or any other cluster.
     130''John Turner'' let's put this in as concrete terms as possible, I
     131have a very difficult time connecting your abstract diagrams with my
     132cluster or any other cluster.
    127133
    128134We need to figure out what people need to do to support this by the fall.
     
    133139goal. Entire cluster is going to be evaluated on getting this to work.
    134140
    135 9:45 Enterprise GENI view of the world, (Rob Sherwood)
     141== Enterprise GENI view ''Rob Sherwood'' ==
    136142
    137143!OpenFlow overview
     
    147153connectivity
    148154
    149 (Chip) have you thought about measurement yet?
     155''Chip Elliott'' have you thought about measurement yet?
    150156
    151157Built into openflow -- byte and packet counters. With a controller you can
    152158redirect flows through a measurement box.
    153159
    154 (Guido) We haven't thought it through in full detail, but you get a fair amount
    155 of control from !OpenFlow, can look deep into a packet.
    156 
    157 We don't really have nodes in a traditional sense, have a datapath ID (i.e.
    158 MAC addr off switch), list of interfaces. We don't "log into" a switch.
    159 
    160 (Guido) as soon as you reserve a switch, the switch connects back to the URL of
    161 your controller and the switch starts asking your controller for
    162 instructions.
    163 
    164 (Camilo Viecco) Do you have one user at a time, or multiple users?
     160''Guido Appenzeller'' We haven't thought it through in full detail, but you get a
     161fair amount of control from !OpenFlow, can look deep into a packet.
     162
     163We don't really have nodes in a traditional sense, have a datapath ID
     164(i.e.  MAC addr off switch), list of interfaces. We don't "log into" a
     165switch.
     166
     167''Guido Appenzeller'' as soon as you reserve a switch, the switch connects back to
     168the URL of your controller and the switch starts asking your
     169controller for instructions.
     170
     171''Camilo Viecco'' Do you have one user at a time, or multiple users?
    165172
    166173You have one user at a time. Default rule is if we don't have a rule for a
    167174packet, a message gets sent to the controller.
    168175
    169 (Guido) If you connect to something that is not part of your aggregate, it's
    170 represented differently. This describes internal references.
     176''Guido Appenzeller'' If you connect to something that is not part of your
     177aggregate, it's represented differently. This describes internal
     178references.
    171179
    172180Can think of !FlowSpace as header, "field=value" pairs plus actions. Packet
     
    176184Ex: all web traffic, except to main server:
    177185
     186{{{
    178187    ip_src = 1.2.3.4 tcp_dport=80 :: DENY
    179 
    180188    ip_src=1.2.3/24 tcp_dport=80 :: ALLOW
    181 
    182 (Guido) can say "ipv6 goes to this controller, ipv4 goes to that controller."
     189}}}
     190
     191''Guido Appenzeller'' can say "ipv6 goes to this controller, ipv4 goes to that controller."
    183192
    184193Rspec - opt-in. How do we express what users experimenters want to allow in?
     
    187196How do we do this between slivers?
    188197
    189 Use case: "gibev me our planetlab nodes and the E-GENI network that connects
     198Use case: "gibv me our !PlanetLab nodes and the E-GENI network that connects
    190199them." Need to know how to communicate that off of this switch, off of this
    191200node, is a point of attachment.
    192201
    193 (Aaron Falk) if you've got multiple slivers on a single planetlab node, how do you
    194 assign them to an egeni node? what does planetlab demultiplex on?
    195 
    196 (Larry Peterson) tcp ports. we've been lazy in how you lock down ports, you claim a port on
    197 a wiki.
    198 
    199 (Aaron Falk) There is a bootstrapping problem wth planetlab and E-GENI. We need to
    200 figure this out.
    201 
    202 (Chip) do you have both openflow and planetlab nodes in the same room?
     202''Aaron Falk'' if you've got multiple slivers on a single planetlab
     203node, how do you assign them to an egeni node? what does planetlab
     204demultiplex on?
     205
     206''Larry Peterson'' tcp ports. we've been lazy in how you lock down
     207ports, you claim a port on a wiki.
     208
     209''Aaron Falk'' There is a bootstrapping problem wth planetlab and
     210E-GENI. We need to figure this out.
     211
     212''Chip Elliott'' do you have both openflow and planetlab nodes in the same room?
    203213
    204214I do not.
    205215
    206 (Ivan Seskar), (Nick Feamster) have both planetlab nodes and openflow
     216''Ivan Seskar'', ''Nick Feamster) have both planetlab nodes and openflow
    207217switches, but they are not connected
    208218
    209 (Larry Peterson) we could have a global allocation of ports, tunnel numbers,
    210 etc., if we just have a global list.
    211 
    212 (Guido) we want a dynamic mapping to slices
    213 
    214 (Ted Faber) If yo're going to define slices in the rspec, have to use globally
    215 understood parts of the flowspec, internally to the aggregate switches may
    216 need to be topoology aware
    217 
    218 10:30 Robert Ricci:  Where We Are
     219''Larry Peterson'' we could have a global allocation of ports, tunnel
     220numbers, etc., if we just have a global list.
     221
     222''Guido Appenzeller'' we want a dynamic mapping to slices
     223
     224''Ted Faber'' If yo're going to define slices in the rspec, have to
     225use globally understood parts of the flowspec, internally to the
     226aggregate switches may need to be topoology aware
     227
     228== ProtoGENI CF View ''Rob Ricci'' ==
    219229
    220230Working prototype rspec
     
    222232Supports nodes, interfaces, links.
    223233
    224 Used to allocate slivers -- raw PCs, vms, vlans, tunnels. Expressed in XML.
     234Used to allocate slivers -- raw PCs, vms, VLANs, tunnels. Expressed in XML.
    225235Tunnels are cross-aggregate. Slice Emnbedding Servive that understands it.
    226236
     
    232242than one.
    233243
    234 Gives to CM, CM signs (generates ticket), '''Manifest''' returned by CM, adds
    235 details like access method, MACs, etc.
     244Gives to CM, CM signs (generates ticket), '''Manifest''' returned by
     245CM, adds details like access method, MACs, etc.
    236246
    237247Four types, similar but not identical.
     
    253263Discussion of how to do what is essentially the travel agent problem.
    254264
    255 [John Duerig]
    256 
    257265Looking at the rspec as a mapping between the requested sliver and the
    258266physical resources.
    259267
    260 (Aaron Falk) What does nick need to do with the bgpmux to use this?
     268''Aaron Falk'' What does nick need to do with the bgpmux to use this?
    261269
    262270We're always adding information, never removing information advertisements
     
    266274A sliver uniquely identiied by (slice ID, virtual ID, CM ID)
    267275
    268 (Aaron Falk) If what I'm advertising is a collection of stuff, what do I advertise?
     276''Aaron Falk'' If what I'm advertising is a collection of stuff, what
     277do I advertise?
    269278
    270279If you don't want to show me the details of your network, it's is not our
    271280design center.
    272281
    273 (Aaron Falk) but I2 won't run an AM, won't identify all the optical switches
    274 along the path
    275 
    276 we'll advertise "here's an enet switch, here's another enet switch", and
     282''Aaron Falk'' but Internet2 won't run an AM, won't identify all the
     283optical switches along the path
     284
     285We'll advertise "here's an enet switch, here's another enet switch", and
    277286won't say anything about the topology beneath it, since it's dynamic and out
    278287of our control.
     
    289298#s, unless translation is possible), Assumption is that cross-aggregate
    290299
    291 (Nick Feamster) VINI has rspec to create tunnels between virtual nodes, but
     300''Nick Feamster'' VINI has rspec to create tunnels between virtual nodes, but
    292301need one to connect VINI to mux, nether VINI or !ProtoGeni
    293302
    294 (Nick Feamster) is there one rspec that's going to say "I need a virtual node
    295 that is a tunnel to this mux"?
    296 
    297 (laughter)
     303''Nick Feamster'' is there one rspec that's going to say "I need a
     304virtual node that is a tunnel to this mux"?
     305
     306''laughter''
    298307
    299308this is all typed, types are well-known device classes (e.g. openflow
     
    303312its links are type "ipv4".
    304313
    305 (Guido) You're assuming these connections are always layer 2? What if it's
     314''Guido Appenzeller'' You're assuming these connections are always layer 2? What if it's
    306315something else?
    307316
     
    313322Links can cross aggregate boundaries -- nodes may not.
    314323
    315 (Ivan Seskar) Said "node cannot cross aggregate" -- but this is common in wireless,
    316 e.g. wifi and wimax.
    317 
    318 (Aaron Falk) Ah, the ''node'' will be in one aggregate, will have two different
    319 kinds of links out (via different carriers, etc.)
     324''Ivan Seskar'' Said "node cannot cross aggregate" -- but this is
     325common in wireless, e.g. wifi and wimax.
     326
     327''Aaron Falk'' Ah, the ''node'' will be in one aggregate, will have
     328two different kinds of links out (via different carriers, etc.)
    320329
    321330Disconnect -- some people think in terms of nodes, some think in terms of
     
    326335a. client negoitates with each CM, rspec is the medium.
    327336
    328 b. cms coordinate among themselves, using a ''new'' standardized control plane
    329 API. Rspec ''could'' be the medium.
     337b. cms coordinate among themselves, using a ''new'' standardized
     338control plane API. Rspec ''could'' be the medium.
    330339
    331340c. Untrusted intermediaty negotiates for client, intermediate has no privs
     
    335344intermediary and CMs. Rspec ''could'' be the medium.
    336345
    337 (Aaron Falk) Would Nick's dynamic tunnel server be an example of (b)?
     346''Aaron Falk'' Would Nick's dynamic tunnel server be an example of (b)?
    338347
    339348Yes
     
    344353oblivious, only CMs talk to intermediary, negotiation information held by CM.
    345354
    346 (Aaron Falk) is this consistent with DRAGON approach?
    347 
    348 (John Tracy) yes, generally -- I've got info in my slides.
    349 
    350 ----------------------------------------------------------------
    351 break
    352 ----------------------------------------------------------------
    353 
    354 Larry Peterson: Resource Specifications and End-to-End Slices
     355''Aaron Falk'' is this consistent with DRAGON approach?
     356
     357''John Tracy'' yes, generally -- I've got info in my slides.
     358
     359== PlanetLab: Resource Specifications / End-to-End Slices ''Larry Peterson'' ==
    355360
    356361This is kind of high level.
    357362
    358 I'm going to argue we have a bunch of nodes. It might be the case that some of
    359 these nodes are special -- e.g., underneath them they have a layer 2 technology they
    360 want to take advantage of.
     363I'm going to argue we have a bunch of nodes. It might be the case that
     364some of these nodes are special -- e.g., underneath them they have a
     365layer 2 technology they want to take advantage of.
    361366
    362367Some of these nodes are going to be part of other aggregates that have special
    363368capabilities, e.g. !OpenFlow aggregate nodes, VINI aggregate nodes.
    364369
    365 (Christopher Small GPO) so a node is a member of an aggregate for each kind of
    366 network it is on?
     370''Christopher Small GPO'' so a node is a member of an aggregate for
     371each kind of network it is on?
    367372
    368373A node is controlled by only one aggregate at a time.
    369374
    370 My definition of a node is something I can dump code into. "Clouds" export an
    371 aggregate manager interface. I can say "set up a circuit between node A and
    372 node B".
     375My definition of a node is something I can dump code into. "Clouds"
     376export an aggregate manager interface. I can say "set up a circuit
     377between node A and node B".
    373378
    374379VINI is a cloud of nodes. Enterprise GENI is a cloud.
    375380
    376 The reason I want to look at the world this way, is that a bunch of nodes
    377 already have a functioning interconnect, the internet. The assumption that I
    378 need across two aggregate boundaries is that I have shared some dmux key
    379 across the boundary, so there's a global allocation of these keys.
     381The reason I want to look at the world this way, is that a bunch of
     382nodes already have a functioning interconnect, the internet. The
     383assumption that I need across two aggregate boundaries is that I have
     384shared some dmux key across the boundary, so there's a global
     385allocation of these keys.
    380386
    381387The world is a ''whole'' lot simpler if everyone is reachable via a shared id
     
    386392this all work more easily.
    387393
    388 (Aaron Falk) You're assuming that there is one of these things between each pair -- what
    389 if you have to go across three links?
    390 
    391 It's complicating life a lot ( for the researcher ) to have to deal with all
    392 pairwise layer 2 possibilities. Give me some guarantees about latency,
    393 failure, bandwidth, independent of what layers of encapsulation I do, is the
    394 key to what the researcher wants to do.
     394''Aaron Falk'' You're assuming that there is one of these things
     395between each pair -- what if you have to go across three links?
     396
     397It's complicating life a lot (for the researcher) to have to deal
     398with all pairwise layer 2 possibilities. Give me some guarantees about
     399latency, failure, bandwidth, independent of what layers of
     400encapsulation I do, is the key to what the researcher wants to do.
    395401
    396402I'm not removing the capability of working with different kinds of links, just
    397403abstracting it away.
    398404
    399 (Ted Faber) Can you view this without having IP tunneling?
     405''Ted Faber'' Can you view this without having IP tunneling?
    400406
    401407I view connecting via something lower than IP that it's enough more difficult
     
    406412Jennifer (who is interested in IP networks) is happy with this. 
    407413
    408 We have built VINI and we have built !PlanetLab. Nobody's coming to VINI to use
    409 layer 2 circuits. They use VINI for the guarantees, not layer 2-level hacking.
    410 
    411 (Rob Ricci) I have a theory that people aren't using VINI for this is because they're
    412 using emulab. A significant minority do experiments ''below'' IP. Playing around
    413 with ethernet bridging, alternatives to IP, ... Still a minority, but we have
    414 them.
     414We have built VINI and we have built !PlanetLab. Nobody's coming to
     415VINI to use layer 2 circuits. They use VINI for the guarantees, not
     416layer 2-level hacking.
     417
     418''Rob Ricci'' I have a theory that people aren't using VINI for this
     419is because they're using emulab. A significant minority do experiments
     420''below'' IP. Playing around with ethernet bridging, alternatives to
     421IP, ... Still a minority, but we have them.
    415422
    416423There are a couple of reasons people aren't using VINI in large numbers, but
    417424there are an awful lot who are most interested in predictable link behavior.
    418425
    419 (Ivan Seskar) Most of the orbit experimenters don't care about IP at all. But that's the
    420 edge.
     426''Ivan Seskar'' Most of the orbit experimenters don't care about IP at
     427all. But that's the edge.
    421428
    422429That's a good point. I'd view ORBIT as another cloud (another aggregate).
    423430
    424 (Ted Faber) If you try to tunnel a MAC over IP, well, it doesn't work very well. Once
    425 you go up to layer 3, you've disrupted layer 2 sufficiently that you can't
    426 necessarily run the experiments you want to run.
     431''Ted Faber'' If you try to tunnel a MAC over IP, well, it doesn't
     432work very well. Once you go up to layer 3, you've disrupted layer 2
     433sufficiently that you can't necessarily run the experiments you want
     434to run.
    427435
    428436As a consequence of this, there may be aggregate-specific experiments.
     
    433441have)
    434442
    435 Have WSDL version of the interface (program-heavy), also XSD version (data heavy)
     443Have WSDL version of the interface (program-heavy), also XSD version
     444(data heavy)
    436445
    437446Backing off of pushing massive nested rspec on you.
     
    439448Adopted simple model:
    440449
     450{{{
    441451        RSpec = !GetResources()
     452}}}
    442453
    443454returns list of all resources available
    444455
     456{{{
    445457        !SetResources(RSpec)
     458}}}
    446459
    447460acquire all resources
     
    449462Only way today is
    450463
     464{{{
    451465     while (true)
    452466       if !SetResources(RSpec)
    453467         break
     468}}}
    454469
    455470Doesn't neessarily terminate.
    456471
    457 Aggregate returns capacity (what it will say yes to in XSD) and policy (how to
    458 interpret the capacity in XSLT). P(Request, Capacity) -> True means request
    459 will be honored. P(Request, Capacity) -> False means request will not be
    460 honored.
    461 Examples:
     472Aggregate returns capacity (what it will say yes to in XSD) and policy
     473(how to interpret the capacity in XSLT). P(Request, Capacity) -> True
     474means request will be honored. P(Request, Capacity) -> False means
     475request will not be honored.  Examples:
    462476
    463477VINI today
     478[[[
    464479     P(R, C) -> true if R and C are the same graph
     480}}}
    465481
    466482VINI tomorrow
     483{{{
    467484     P(R, C) -> true if R is a subset of C
     485}}}
    468486
    469487!PlanetLab today
     488{{{
    470489          P(R, C) -> true if R is a subset of C and site sliver count OK
    471 
    472 (Nick Feamster) Is there a notion of time in an rspec?
     490}}}
     491
     492''Nick Feamster'' Is there a notion of time in an rspec?
    473493
    474494Yes
     
    476496Discussion of using pyton vs XSLT for this.
    477497
    478 (Aaron Falk) We're off track, you've gotten us off into rspec reconcilliation.
    479 
    480 ----------------------------------------------------------------
    481 
    482 Lunch
    483 
    484 ----------------------------------------------------------------
    485 
    486 Ilya: experimenting with ontologies for multi-layer network slicing
     498''Aaron Falk'' We're off track, you've gotten us off into rspec
     499reconcilliation. 
     500
     501== ORCA, BEN, NDL-OWL ''Ilya Baldine'' ==
     502
     503Ilya Baldine: experimenting with ontologies for multi-layer network slicing
    487504
    488505Need a way to describe what we have (substrate), what we want (request), what
     
    498515
    499516we aren't the first to face this: netowkr marjup language working group
    500 (NML-WG). Participants include I2, ESnet (!PerfSONAR model) Dante/ GN2.
     517(NML-WG). Participants include Internet2, ESnet (!PerfSONAR model) Dante/ GN2.
    501518University Amsterdam (NDL)
    502519
     
    522539Goals -- more description languages, meaurement, cloud, wireless, etc.
    523540
    524 (Ted Faber) My concern is that it seems very detailed. More detail than we need?
    525 
    526 (Ivan Seskar) Example: give me a linear topology of nodes
    527 
    528 (Aaron Falk) Assume there are tools that translate high-level descr into this.
     541''Ted Faber'' My concern is that it seems very detailed. More detail than we need?
     542
     543''Ivan Seskar'' Example: give me a linear topology of nodes
     544
     545''Aaron Falk'' Assume there are tools that translate high-level descr into this.
    529546
    530547I don't think people will point and click their way to this.
    531548
    532 (Chip) Glyph could be federated into GENI and vice versa
     549''Chip Elliott'' Glyph could be federated into GENI and vice versa
    533550
    534551We're working on it.
    535552
    536 ----------------------------------------------------------------
    537 1:15pm
    538 ----------------------------------------------------------------
    539 
    540 MAX/DRAGON view, Chris Tracy
     553== MAX/DRAGON ''Chris Tracy'' ==
    541554
    542555SOAP-based GENI aggregate manager.
     
    547560an apache service, uses WSDL (web services API).
    548561
    549 (Larry Peterson) We have a SOAP interface now, too, should be able to
     562''Larry Peterson'' We have a SOAP interface now, too, should be able to
    550563interoperate.
    551564
     
    553566!PlanetLab controller.
    554567
    555 (Chip) is !OpenFlow currently using same or different SOAP interface?
    556 
    557 (Larry Peterson) It's a subset, we need to have the discussion and get them in sync
     568''Chip Elliott'' is !OpenFlow currently using same or different SOAP interface?
     569
     570''Larry Peterson'' It's a subset, we need to have the discussion and get them in sync
    558571
    559572We've mostly tried to stick with what was in the slice facility architecture
     
    568581We want this aggregate manager to be able ot manage anything on the net.
    569582
    570 (Aaron Falk) This aggregate manager box isn't just a bunch of functions, doing some work
    571 to make sure things are allocated in a coherent manner
    572 
    573 You can go to this AM and run "list capabilities" or "get nodes" and pass in a
    574 filter spec (give me all the nodes that can do both dragon and planetlab).
    575 Returns a controller URL so you can go talk to the controller for more info.
     583''Aaron Falk'' This aggregate manager box isn't just a bunch of
     584functions, doing some work to make sure things are allocated in a
     585coherent manner
     586
     587You can go to this AM and run "list capabilities" or "get nodes" and
     588pass in a filter spec (give me all the nodes that can do both dragon
     589and planetlab).  Returns a controller URL so you can go talk to the
     590controller for more info.
    576591
    577592Code is published on the website, instances will be site-specific (aggregate
     
    580595Wrote WSDL file by hand based on SFA. wsdl2java generated Java skeleton code.
    581596
    582 (http://geni.dragon.maxgigapop.net:8080/axis2/services/AggregateGENI?wsdl demo
    583 using a generic SOAP Client)
    584 
    585 (Chip) Nick is this the AM you should be using?
    586 
    587 (Rob Ricci) in our case we haven't described our interface as a WSDL
    588 
    589 (Rob Sherwood) there are a lot of WSDL tools you can use
     597(Opens
     598http://geni.dragon.maxgigapop.net:8080/axis2/services/AggregateGENI?wsdl
     599demo using a generic SOAP Client)
     600
     601''Chip Elliott'' Nick is this the AM you should be using?
     602
     603''Rob Ricci'' in our case we haven't described our interface as a WSDL
     604
     605''Rob Sherwood'' there are a lot of WSDL tools you can use
    590606
    591607The code for this (svn repo) is pointed to in the slides (p. 11?)
     
    617633layer 2, not cross-layer allocation.
    618634
    619 (Chip) the GPO architecture would have a central clearinghouse, messages going up
    620 and down; in this the messages go across.
     635''Chip Elliott'' the GPO architecture would have a central clearinghouse,
     636messages going up and down; in this the messages go across.
    621637
    622638(OGF26 presentation NDL working group, Multilayer NDL presentation by Freek
     
    625641This is GMPLS inspired, done with signaling via web services.
    626642
    627 (Yufeng Xin) Who issues cross boundary configurations?
     643''Yufeng Xin'' Who issues cross boundary configurations?
    628644
    629645Once there's agreement over which VLAN we're terminating on, each aggregate
    630646will do it.
    631647
    632 (Chip) how baked is this? Used 10 times a day?
     648''Chip Elliott'' how baked is this? Used 10 times a day?
    633649
    634650Hundreds of times a month. Solid. Pretty much always works.
    635651
    636 ----------------------------------------------------------------
    637 Break
    638 ----------------------------------------------------------------
    639 Aaron
     652== Planning ''Aaron Falk'' ==
    640653
    641654What will each cluster be doing to reach the goal of cross-aggregate slices by
    642655the end of spiral 1? What are the inter-project dependencies?
    643656
    644 (Larry Peterson) you're forcing us into realtime project negotiation here. I
    645 think we ought to do as much as we can assuming IP as the interconnect. It
    646 works for some, maybe not all -- !GpENI? SPP boxes?
    647 
    648 (John Turner) users can log into each and allocate by hand. No more explicit
     657''Larry Peterson'' you're forcing us into realtime project negotiation
     658here. I think we ought to do as much as we can assuming IP as the
     659interconnect. It works for some, maybe not all -- !GpENI? SPP boxes?
     660
     661''John Turner'' users can log into each and allocate by hand. No more explicit
    649662coordination is required to make it work.
    650663
    651664What's needed to get there from here?
    652665
    653 (James Sterbenz) From our perspective stitching together with IP works for
     666''James Sterbenz'' From our perspective stitching together with IP works for
    654667now, but long term for GENI to succeed need to support more.
    655668
    656669Is doing this by hand workable? Does this work for everyone?
    657670
    658 (Chris Tracy) We can provision nodes via DRAGON
    659 
    660 (John Turner) For both !GpENI and DRAGON we can terminate a connection that we
    661 have VLANs defined on. There is a non-trivial amount of control software that
    662 we need to write but we have other things to do first, like getting systems
    663 deployed.
     671''Chris Tracy'' We can provision nodes via DRAGON
     672
     673''John Turner'' For both !GpENI and DRAGON we can terminate a
     674connection that we have VLANs defined on. There is a non-trivial
     675amount of control software that we need to write but we have other
     676things to do first, like getting systems deployed.
    664677
    665678Goal is constructing end-to-end slices, not having it done automagically.
     
    668681stanford campus.
    669682
    670 (Guido) I think IP is a good common denominator for connecting aggregates for
     683''Guido Appenzeller'' I think IP is a good common denominator for connecting aggregates for
    671684now. If we want to scale this to hundreds of aggregates this won't work.
    672685
     
    675688connections.
    676689
    677 (John Turner) Did a version of this for GEC3.
     690''John Turner'' Did a version of this for GEC3.
    678691
    679692Yes, but we need this for the cluster. Nobody has put onto a single sheet of
    680693paper all the things that need to be done to do this.
    681694
    682 (John Turner) We all connect to the outside world via Internet2.
    683 
    684 We have our own wave on I2. There are not routers on it. Want to draw a
     695''John Turner'' We all connect to the outside world via Internet2.
     696
     697We have our own wave on Internet2. There are not routers on it. Want to draw a
    685698distinction between our access-to-the-outside-world and the GENI backbone.
    686699
    687700Goal is to demonstrate end-to-end slices across a range of technologies.
    688701
    689 (Chris Tracy) Are you in DC yet?
    690 
    691 (Robert Ricci) There will be within a small number of weeks.
    692 
    693 Action: Chris Tracy will get into the Internet2 cage, and will pull a cable
    694 between DRAGON and SPP (?).
     702''Chris Tracy'' Are you in DC yet?
     703
     704''Rob Ricci'' There will be within a small number of weeks.
     705
     706'''Action: Chris Tracy will get into the Internet2 cage, and will pull a cable
     707between DRAGON and SPP (?).'''
    695708
    696709Internet2 has told us we may be able to get access to get people from DCN to
    697710the GENI wave.
    698711
    699 (Rob Sherwood) What are the right interfaces? We are in LA, Houston, and NY,
     712''Rob Sherwood'' What are the right interfaces? We are in LA, Houston, and NY,
    700713are adding DC.
    701714
    702 (Chris Tracy) How is !GpENI going to connect to Internet2 DCN or the GENI wave?
    703 
    704 (James Sterbenz) to maxgigapop, our equipment is in the internet2 pop in
     715''Chris Tracy'' How is !GpENI going to connect to Internet2 DCN or the
     716GENI wave?
     717
     718''James Sterbenz'' to maxgigapop, our equipment is in the Internet2 pop in
    705719Kansas City.
    706720
    707 (Chip) How much will this cost?
    708 
    709 (James Sterbenz) Will take action item to find out how much this will cost.
    710 
    711 (Robert Ricci) You need to determine this quickly, we have a switch going in
     721''Chip Elliott'' How much will this cost?
     722
     723''James Sterbenz'' Will take action item to find out how much this will cost.
     724
     725''Rob Ricci'' You need to determine this quickly, we have a switch going in
    712726in the next couple of weeks. We need to talk.
    713727
    714 Action: Rob, James, Heidi coordinate on Kansas City Internet2 connection
    715 
    716 Action: Aaron will send email to Rob and James to make sure they can contact
    717 each other via email.
     728'''Action: Rob, James, Heidi coordinate on Kansas City Internet2 connection'''
     729
     730'''Action: Aaron Falk will send email to Rob and James to make sure
     731they can contact each other via email.'''
    718732
    719733MAX connects to NLR.
    720734
    721 Action: Guido, John Turner will write a one-page high level list of the
    722 actions one needs to take to configure a slice.
    723 
    724 (Robert Ricci) We're already doing this, more or less. It's done with
     735'''Action: Guido Appenzeller, John Turner will write a one-page high
     736level list of the actions one needs to take to configure a slice.'''
     737
     738''Rob Ricci'' We're already doing this, more or less. It's done with
    725739tunnels. Once we get set up in Internet2 POPs. Plan outlined on his last slide
    726740shows how to get VLANs from campus to campus.
    727741
    728 (Robert Ricci) Kentucky, CMU are already in. UML PEN shouldn't be too hard.
     742''Rob Ricci'' Kentucky, CMU are already in. UML PEN shouldn't be too hard.
    729743
    730744The picture will be very helpful
    731745
    732 (Robert Ricci) It was on my poster at the last GEC.
    733 
    734 Action: Robert Ricci will pull together the picture.
     746''Rob Ricci'' It was on my poster at the last GEC.
     747
     748'''Action: Rob Ricci will pull together the picture.'''
    735749
    736750Cluster D, the impression that I've got is that you're all pretty integrated
     
    738752to BEN?
    739753
    740 (Ilya) Technically we know some of the problems are. GPO committed us to being
     754''Ilya Baldine'' Technically we know some of the problems are. GPO committed us to being
    741755an NLR based cluster. UMass is working to get sone VLANs, but they are a
    742756limited resource.  We are trying to figure out how to get to Charlotte
    743 (Internet2 terminates there) from RTP.  Maybe MPLS or VLAN from RENCI BEN POP
     757''Internet2 terminates there) from RTP.  Maybe MPLS or VLAN from RENCI BEN POP
    744758to Intenet2.
    745759
     
    751765Let's get this picture so we can figure out where the gaps are.
    752766
    753 (Ilya) My main problem is that there will be costs associated with connecting
    754 us to Internet2. We don't know how much.
    755 
    756 (Harry) Does it make sense to draw a picture of BEN, NLR, and MAX?
    757 
    758 (Ilya) I'd like to do this, let's talk about this.
    759 
    760 Action: Harry and Ilya will talk about this.
     767''Ilya Baldine'' My main problem is that there will be costs associated with
     768connecting us to Internet2. We don't know how much.
     769
     770''Harry Mussman'' Does it make sense to draw a picture of BEN, NLR, and MAX?
     771
     772''Ilya Baldine'' I'd like to do this, let's talk about this.
     773
     774'''Action: Harry and Ilya Baldine will talk about this.'''
    761775
    762776Cluster E?
    763777
    764 (Ivan Seskar) Hey, we're done. We're on the same campus. Except for the air
    765 gap; we need to get someone to pull a cable up six floors from where Internet2
    766 terminates and where we are.
     778''Ivan Seskar'' Hey, we're done. We're on the same campus. Except for
     779the air gap; we need to get someone to pull a cable up six floors from
     780where Internet2 terminates and where we are.
    767781
    768782Cluster A?
    769783
    770 (Ted Faber) We're trying to do some relevant end-to-end work. Hook up to the
    771 DCN, plugged into a DETERLab node on one end, ISI East on the other. Working
    772 on the expanded authorization work we have talked about at the last couple of
    773 GECs.
     784''Ted Faber'' We're trying to do some relevant end-to-end work. Hook
     785up to the DCN, plugged into a DETERLab node on one end, ISI East on
     786the other. Working on the expanded authorization work we have talked
     787about at the last couple of GECs.