Changes between Version 2 and Version 3 of _GIMI_v1.0_Architecture


Ignore:
Timestamp:
05/25/12 14:28:49 (9 years ago)
Author:
hmussman@bbn.com
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • _GIMI_v1.0_Architecture

    v2 v3  
    11[[PageOutline]]
    22
     3
    34== GIMI v1.0 Architecture for GEC 14 Tutorial  ==
    4 
    5 GIMI v1.0 Architecture for GEC 14 Tutorial  [[BR]]
    65
    76This document gives an overview of the GIMI v1.0 architecture as it is planned to be in place for the tutorial at GEC 14.  [[BR]]
     
    98After review on 5/23/12.  [[BR]]
    109
    11 General  [[BR]]
     10=== General  ===
    1211
    1312 The tutorial is limited to a maximum of 30 attendants. Two attendees will be working as a team during the tutorial.  Like the DICloud tutorial, there will be several members of the GIMI team there to allow us to pay sufficient attention to each individual.
     
    2928
    3029
    31 ExoGENI  [[BR]]
    32  •      For the tutorial users will be given a pre-defined ExoGENI topology. Some of these topologies (approximately 3-4) will be inter rack slices (two aggregates), the others will be intra rack slices (one aggregate).
    33  •      We will give a short (approximately 30 minutes) introduction on how resources can be obtained with either Flukes or Omni.  This will be canceled if an ExoGENI tutorial precedes our tutorial at GEC14.
    34  •      Due to time limit we will use per-configured ExoGENI machine images. Users will have to learn otherwise how to generate and configure images.
    35  •      The pre-configured images will include (5/25):
    36   o     OML 2.7.0
    37   o     OMF 2.7.0
    38   o     OML-ified Iperf
    39   o     iRODS 3.1 client
    40   o     web server  (small apache)
    41   o     r (for visualization)
    42   o     also EC?
     30=== ExoGENI  ===
     31 •      For the tutorial users will be given a pre-defined ExoGENI topology. Some of these topologies (approximately 3-4) will be inter rack slices (two aggregates), the others will be intra rack slices (one aggregate). [[BR]]
     32 •      We will give a short (approximately 30 minutes) introduction on how resources can be obtained with either Flukes or Omni.  This will be canceled if an ExoGENI tutorial precedes our tutorial at GEC14. [[BR]]
     33 •      Due to time limit we will use per-configured ExoGENI machine images. Users will have to learn otherwise how to generate and configure images. [[BR]]
     34 •      The pre-configured images will include (5/25): [[BR]]
     35  o     OML 2.7.0 [[BR]]
     36  o     OMF 2.7.0 [[BR]]
     37  o     OML-ified Iperf [[BR]]
     38  o     iRODS 3.1 client [[BR]]
     39  o     web server  (small apache) [[BR]]
     40  o     r (for visualization) [[BR]]
     41  o     also EC? [[BR]]
    4342
    4443To do:  [[BR]]
    45  •      Review GEC 13 ExoGENI tutorial slides.
    46  •      Coordinate with GEC14 ExoGENI tutorial.
    47  •      Create accounts that can be used to access ExoGENI resources (6/15)
    48  •      Pre-configure 15 ExoGENI slices on aggregates (define rspecs) (6/15)
     44 •      Review GEC 13 ExoGENI tutorial slides. [[BR]]
     45 •      Coordinate with GEC14 ExoGENI tutorial. [[BR]]
     46 •      Create accounts that can be used to access ExoGENI resources (6/15) [[BR]]
     47 •      Pre-configure 15 ExoGENI slices on aggregates (define rspecs) (6/15) [[BR]]
    4948 
    50 OMF/OML  [[BR]]
    51  •      Run OML server on one of the ExoGENI nodes (5/25).
    52  •      Use OML-ified Iperf for throughput measurements between ExoGENI nodes (5/25)
    53  •      Use nmetrics to monitor condition of single node (5/25)
    54  •      OMF EC to execute measurement (5/25)
    55  •      Collect data on MC (5/25)
     49=== OMF/OML ===
     50 •      Run OML server on one of the ExoGENI nodes (5/25). [[BR]]
     51 •      Use OML-ified Iperf for throughput measurements between ExoGENI nodes (5/25) [[BR]]
     52 •      Use nmetrics to monitor condition of single node (5/25) [[BR]]
     53 •      OMF EC to execute measurement (5/25) [[BR]]
     54 •      Collect data on MC (5/25) [[BR]]
    5655
    5756To do:  [[BR]]
    58  •      How to bring up desired services on selected nodes?  (6/1)
    59  •      Decide on what XMPP server to use. NICTA, Rutgers, UMass? (5/25)
    60  •      Decide how to load info into XMPP server.  (6/1)
    61  •      Provide for visualization on MC (6/1)  How?
     57 •      How to bring up desired services on selected nodes?  (6/1) [[BR]]
     58 •      Decide on what XMPP server to use. NICTA, Rutgers, UMass? (5/25) [[BR]]
     59 •      Decide how to load info into XMPP server.  (6/1) [[BR]]
     60 •      Provide for visualization on MC (6/1)  How? [[BR]]
    6261 
    63 iRODS  [[BR]]
    64  •      Persistent iRODS servers are already setup at UMass and RENCI (with RENCI running iCAT).
    65  •      Measurement data collected at MC will be transferred to iRODS (6/1).
    66  •      Users will be able to use IREEL to analyze and present data stored in iRODS (6/15).
     62=== iRODS  ===
     63 •      Persistent iRODS servers are already setup at UMass and RENCI (with RENCI running iCAT). [[BR]]
     64 •      Measurement data collected at MC will be transferred to iRODS (6/1). [[BR]]
     65 •      Users will be able to use IREEL to analyze and present data stored in iRODS (6/15). [[BR]]
    6766
    6867To Do:  [[BR]]
    69  •      Create iRODS accounts for tutorial users (6/1)
    70  •      We should give a short (approximately 30 minutes) iRODS intro during the tutorial. MZ: Still possible with in 90 minutes time slot?
     68 •      Create iRODS accounts for tutorial users (6/1) [[BR]]
     69 •      We should give a short (approximately 30 minutes) iRODS intro during the tutorial. MZ: Still possible with in 90 minutes time slot? [[BR]]
    7170
    72  IREEL  [[BR]]
    73  •      Use IREEL as measurement portal (6/1)
    74  •      Persistent portal will be set up at UMass Amherst (6/15)
     71 === IREEL  ===
     72 •      Use IREEL as measurement portal (6/1) [[BR]]
     73 •      Persistent portal will be set up at UMass Amherst (6/15) [[BR]]
    7574
    7675
    7776To Do:  [[BR]]
    78  •      Implement Flukes/Omni interface (TBD) MZ: Do we want to do this for tutorial?
     77 •      Implement Flukes/Omni interface (TBD) MZ: Do we want to do this for tutorial? [[BR]]
    7978 •      Implement iRODS interface (6/15)
    80  •      Can GENI credential mechanism be used? – Most likely not for tutorial!
     79 •      Can GENI credential mechanism be used? – Most likely not for tutorial! [[BR]]
    8180
    82  VirtualBox  [[BR]]
    83  •      It would be great if we could have an independent (of other tutorials) VirtualBox image;  on 5/23, it appears that only GIMI and GEMINI require VirtualBox, so perhaps have one image for both.
    84  •      VirtualBox image for testing (6/1, Jeanne)
    85  •      Required SW:
    86   o     Flukes
    87   o     Omni
    88   o     iRODS client
    89   o     Web browser
    90   o     EC
     81 === VirtualBox  ===
     82 •      It would be great if we could have an independent (of other tutorials) VirtualBox image;  on 5/23, it appears that only GIMI and GEMINI require VirtualBox, so perhaps have one image for both. [[BR]]
     83 •      VirtualBox image for testing (6/1, Jeanne) [[BR]]
     84 •      Required SW: [[BR]]
     85  o     Flukes [[BR]]
     86  o     Omni [[BR]]
     87  o     iRODS client [[BR]]
     88  o     Web browser [[BR]]
     89  o     EC [[BR]]
    9190
    9291To Do:  [[BR]]
    93  •      Is there a way to pre-install all required credentials?
    94  •      What Java version does Flukes require?
     92 •      Is there a way to pre-install all required credentials? [[BR]]
     93 •      What Java version does Flukes require? [[BR]]
    9594
    96  Applications  [[BR]]
    97  •      Time-series radar data:  measure throughput of time-series data?
    98  •      Video:  measure jitter of video data?
    99  •      TCP/UDP experiment from IREEL, Guillaume?
    100  •      Multi-path TCP?
    101  •      Energy consumption?
     95 === Applications  ===
     96 •      Time-series radar data:  measure throughput of time-series data? [[BR]]
     97 •      Video:  measure jitter of video data? [[BR]]
     98 •      TCP/UDP experiment from IREEL, Guillaume? [[BR]]
     99 •      Multi-path TCP? [[BR]]
     100 •      Energy consumption? [[BR]]
    102101
    103102To Do:  [[BR]]
    104  •      Pick a good application, or “reference experiment”  (6/1)
    105  •      Prepare a reference use case  (6/1)
     103 •      Pick a good application, or “reference experiment”  (6/1) [[BR]]
     104 •      Prepare a reference use case  (6/1) [[BR]]