Anuket Project

2021-11-19 AI/ML for NFV Meeting Minutes

Attendees

Sridhar Rao

Girish

renukananda td

Vidyashree K S

Hemashree R

Ildiko



Sl. No.TopicPresenterNotes
1About Thoth

Updates to the slides presented last week.

Latest version can be found here: About Thoth 

Slide about Nouns and Verbs of Thoth.

2Demo of Synthetic Monitoring/Logging Data Generation using GANs.

Checked with

  1. Barometer project too - They too don't have data sample
  2. Linux Foundation testbeds. They mostly run Baremetal as service. In this scenario, the cloud-provider (the labs) do not run any observability tools.
  3. China Mobile - Considering.

Current: Performance is poor (55%). Working towards improving the same.

Future: Try Table GANs.

ToDo:

  1. Finalize the dataset
    1. Consideration: Multiple sources or Single Source
    2. All infra-dataset is multiple sources.
  2. On Openstack or K8S - Run Functest including performance tests. While running functest we can collect the infrastructures.
3ML in Cloud-Native NFVWork in Progress - Move to Next week for discussion on this.
4MaaSSridhar Rao
  1. Framework - Acumos|Kubeflow.
    1. Deployed Kubeflow, and a pipeline with basic linear regression was tried.
    2. Trying to answer: How easy it is to add newer models - and each model as a separate container.
  2. Human Resource - Researchers - Team
    1. Karthik and Atif. 
    2. Access to Pod18 - for running Kubeflow.
  3. Data Confidentiality Promise.
  4. PoC -  ??? (Problem for which they (end-users) don't want to invest internal Human resources).