wiki:Topic1

Version 1 (modified by hmussman@bbn.com, 13 years ago) (diff)

--

Topic 1 GENI I&M Use Cases

Team members:

Paul Barford - University of Wisconsin – Madison (no)
Jim Griffioen - Univ Kentucky (yes)
Prasad Calyam* - Ohio Supercomputing Ctr (yes)
Camilo Viecco - Indiana Univ (yes)
Brian Hay – Univ Alaska (yes)

*agreed to organize first writing and discussion

To ensure that the GENI I&M architecture serves the needs of a wide range of users, a list of user groups and their related use cases are compiled in this Section.
The user groups identified are:

  1. Experiment Researchers: Users that run Internet-scale experiments in slices comprising of multiple GENI resources to address research problems of the future-Internet.
  2. Experiment (Opt-In) Users: Users within the GENI suite or general Internet users who participate or “opt-in” to a GENI experiment to utilize resources, applications or services that are hosted within the GENI experiment.
  3. Central (i.e., GMOC) Operators: Group that monitors the GENI facility resources and processes in order to bring consistency, reliability and repeatability to GENI’s federated infrastructure.
  4. Aggregate Providers and Operators: Groups that provide a set of network or computing components to GENI experiments along with software to manage the components, and allow users to check the availability and status of the various components.
  5. Archive Providers and Operators: Groups that catalog indexes of GENI-related measurement datasets in a repository and provide tools for users to share, annotate, search and cite the measurement datasets.
  6. Researchers that use Archived Measurement Data: Users that utilize the measurement datasets provided by Archive Providers in order to test hypotheses, and promote reproducible research.

The interfaces to the GENI I&M services that will drive the user actions in the use cases will be either: (i) web-page based, (ii) command-line based or (iii) custom client software based.
The overall goals for the design of the use cases provided in the following sub-sections are for the GENI I&M system (GIMS) to:

  • Provide broad data gathering, analysis and archival capability that is sufficient for scientific mission, operations, and success of the infrastructure
  • Remove burden on researcher to become a system and network measurement infrastructure expert so that researcher can better focus on the science in the experiments
  • Measure details of GENI behavior with high precision and accuracy in a ubiquitous, extensible, highly available, secure, and integrated manner without adversely impacting experiments
  • Provide drill-down performance transparency of system and network resources at hop, link, path and slice levels in terms of availability, health status, and diagnosis of perceived as well as impending problems
  • Allow and make-it-easy for various user groups to access and control functions involving interactions between I & M sub-services encompassing resources such as instrumentation taps in the network, time sensors, software-based and hardware-based measurement probes, router/switch MIBs, and (short-term/long-term) measurement data archives
  • Provide performance transparency of the status of the individual I & M sub-service components and their interfaces with other sub-services to ensure correctness of measurements provisioned
  • Provide mechanisms to handle security, privacy and access control of measurement data archives to allow access only to authorized users, and also provide different data views based on authorization privileges

i. For Experiment Researchers

  • A slice has been setup for me, have I got all the resources with the performance expectations that I specified in the RSpec? For example, I asked for a 2 Mbps available bandwidth connection between Nodes A and B, run a 2 Mbps UDP Iperf test so that I can check there is no packet loss
  • Show me a dashboard of some or all of the resource performance measurements as I run my experiments so that I can have knowledge of my experiment environment in real-time. Allow me to configure the dashboard such that it will be obvious for me to see any impending or perceived problems when measurement values cross my pre-set performance thresholds
  • My experiment data shows inconsistencies, let me query the status of user slice resources so that I can trace my non-intuitive results to a problem in the environment and subsequently notify GMOC about any perceived performance problems
  • Provide me with an archive of some or all of the slice resource performance measurements so that I can reference them during offline analysis of the data collected in my experiment after the slice expires
  • Setup up TCPdump passive measurement taps at hops a, b, c and provide an interface where I can view and analyze the slice components and slivers performance in an on-going manner
  • Setup Netflow measurements collection at hops a, b, c and provide an interface where I can view and analyze the flows in my slice in an on-going manner
  • Setup vendor-specific measurements collection from equipment at hops a, b, c in my slice and provide an interface where I can view and analyze the measurements in an on-going manner
  • Setup up active measurement capabilities on paths x, y, z using p, q, r tools; Provide capabilities for on-demand measurement with quick response times; Provide capabilities for on-going measurements with sampling patterns in {periodic, random, stratified random, adaptive}
  • Setup one-way delay active measurements with microsecond precision on paths x, y, and z encompassing hops a, b, and c that have Netflow measurements collection enabled so that I can know when there is short-term buildup in router queues that does not show in link utilization data, but is correlated with large flows that appear in the flow data
  • I am writing an event-driven experiment, at certain time points, I would like to be notified of anomalies and forecasts of system and network performance at hops a, b, c on paths x, y, z pertaining to tools p, q, r
  • I am running an experiment to deploy a novel IPTV system protocol, provide me with PSNR measurements of video quality between paths x, y, z (e.g., Evalvid tool that will need source and destination packet captures)
  • Provide access to my opt-in users who want to query measurement data within my experiment slice using web-service clients based on GIMA compliant data sharing schemas
  • Provide me with an archive of some or all of the slice resource performance measurements that I requested as part of my experiment
  • Provide me with mechanisms to share my slice measurements archive with researchers and opt-in users at different levels of permissions sharing (i.e., whitelist/blacklist, sign-in, public)

ii. For Experiment (Opt-In) Users

  • I am utilizing a new P2P networking service in a GENI experiment slice, show me the end-to-end delay and loss characteristics of the network paths between my computer and all the P2P servers in the GENI experiment
  • I have subscribed for a virtual desktop service in a GENI experiment slice, show me whether I got all the resources (e.g., CPU, Memory, Disk space) with the performance expectations that I requested in my virtual desktop computer?
  • My application running in the GENI experiment has poor performance, let me query the latest status of my application resources so that I can know the reason for the poor performance or I can notify the researcher who is running the GENI experiment

iii. For Central (i.e., GMOC) Operators

  • For a physical topology of Nodes {A, … Z} spanning multiple aggregates, show me if any slice is mis-behaving so that I can invoke “emergency shutdown” to swap it out
  • Experimenter called NOC about non-responsiveness of resources or unexpected behavior in a slice spanning multiple aggregates, notify status of user slice resources via a dashboard with some or all of the resource performance measurements in the user slice; Allow me to configure the dashboard such that it will be obvious for me to see any impending or perceived problems when measurement values cross pre-set performance thresholds
  • We would like to keep meta-data of all the experiments, send us experiment meta-data after each slice expires
  • Setup Netflow measurements collection at hops a, b, c spanning multiple aggregates and provide an interface where I can view and analyze the flows of all the experiment slices in an on-going manner
  • Setup vendor-specific measurements collection from equipment at hops a, b, c spanning multiple aggregates and provide an interface where I can view and analyze the measurements in an on-going manner
  • Setup up active measurement capabilities on paths x, y, z using p, q, r tools spanning multiple aggregates; Provide capabilities for looking at the measurements being collected via a weathermap interface
  • Provide me with an archive of some or all of the slice resource performance measurements of users X and Y so that I can analyze infrastructure problems spanning multiple aggregates that may have corrupted the users experiment environments

iv. For Aggregate Providers and Operators

  • I would like to have an authentication mechanism for NOC staff, researchers, and opt-in users so that measurement data access spanning across my aggregate components can be granted based on the privileges assigned to the different user roles
  • For a physical topology of Nodes {A, … Z} in my aggregate, show me if any slice is mis-behaving so that I can swap the experiment out and/or reallocate resources
  • The NOC has complained about non-responsiveness of resources or unexpected behavior in a slice using my aggregate, notify status of the user slice resources via a dashboard with some or all of the resource performance measurements in the user slice; Allow me to configure the dashboard such that it will be obvious for me to see any impending or perceived problems when measurement values cross pre-set performance thresholds
  • I would like to keep meta-data of all the running/expired experiments using my aggregate, so that I can track the resource utilization levels and the inherent experiment purposes over time
  • Setup Netflow measurements collection at hops a, b, c in my aggregate and provide an interface where I can view and analyze the flows of all the experiment slices in an on-going manner
  • Setup vendor-specific measurements collection from equipment at hops a, b, c in my aggregate and provide an interface where I can view and analyze the measurements in an on-going manner
  • Setup up active measurement capabilities on paths x, y, z using p, q, r tools in my aggregate; Provide capabilities for looking at the measurements being collected via a weathermap interface
  • Provide me with an archive of some or all of the slice resource performance measurements of users X and Y that are using my aggregate so that I can analyze infrastructure problems that may have corrupted my aggregate users experiment environments

v. For Archive Providers and Operators

  • I would like measurement archives corresponding to GENI experiments to be published in the repositories PQR by the experiment researchers, aggregate providers and GMOC with suitable keywords that allow me to catalog indexes for future search and retrieval purposes
  • I would like to have an authentication mechanism for NOC staff, aggregate providers, experiment researchers, opt-in users, and researchers that use archived measurement data so that measurement data access in my repositories can be granted based on the privileges assigned to the different user roles
  • I would like NOC staff, aggregate providers, and experiment researchers to provide me policies relating to the measurement archive sharing permissions (i.e., whitelist/blacklist, sign-in, public)
  • I would like users to use my tools and transformation libraries that deal with various data formats to: share, annotate, search and cite the measurement datasets in my repositories
  • I would like NOC staff, aggregate providers, and experiment researchers to contribute various tools that will allow researchers using the archived measurement data to analyze and visualize their corresponding published data sets more effectively

vi. For Researchers that use Archived Measurement Data

  • I would like get search results and access to measurement archives corresponding to GENI experiments published by the experiment researchers, aggregate providers and GMOC when I use different search keywords
  • I would like to be able to share (e.g., email, post on Twitter), annotate, search and cite the measurement datasets in repositories of several Archive Providers