KubeCon Shanghai 2019 demo preparation

[xing-yang]

I'm opening this issue to track demo preparation for KubeCon Shanghai which happens on June 24, 2019. See details in the following Google Doc:

https://docs.google.com/document/d/16ZtxByFkbfQyvvF8b8o3IGzWAlJt7pMPPm70YFjR7gw/edit?usp=sharing

@xiangrumei @wisererik PTAL.

[xing-yang]

OpenSDS Capri Demos

KubeCon Shanghai (June 24-26, 2019)
https://www.lfasiallc.com/events/kubecon-cloudnativecon-china-2019/
OpenSDS mini-summit June 24

Telemetry & Anomaly Detection

Prakash, Naju, and Jiajun

Demo setup:
Setup an environment with Kubernetes, OpenSDS CSI plugin with Ceph backend, and MongoDB application running. MongoDB should be using a volume provisioned by OpenSDS CSI plugin on Ceph (Done). Use an MongoDB exporter to collect some metrics on DB performance.
Collect Metrics from the following sources:

• TODO: Performance metrics from MongoDB exporter: https://github.com/dcu/mongodb_exporter
• Metrics from Node-Exporter: https://github.com/prometheus/node_exporter
• Pool level Metrics from Ceph backend
  We don't have volume level performance metrics for Ceph: https://github.com/opensds/opensds/blob/development/contrib/drivers/ceph/ceph_metrics.go#L23.

Note that we should collect all available Ceph metrics, not just pool level, and show them in dashboard. For anomaly detection, we’ll just use pool metrics to run algorithms.

Method to collect metrics:

• Use Data Generator to collect metrics.
• If for some reason option 1 does not work, we can collect metrics manually as before: run scripts for a 2 day period, with 10 sec interval. Simulate stress, use DD and Bonnie++, etc.

Anomaly detection:
Use metrics collected from DB application, node-exporter, and metrics from Ceph to run algorithms (DBSCAN and Gaussian)

Run separately with data from different sources: one run for MongoDB metrics, one run for node-exporter data, and one run for metrics from Ceph, but data MUST have SAME timestamps in the 3 runs so we can show resulting figures side by side.
For MongoDB metrics, pick 1 to 2 metrics to run the algorithms.
For node-exporter, pick 1 to 2 metrics to run the algorithms.

• Screenshots of dashboard to show all types of Ceph metrics collected:
  https://github.com/opensds/opensds/blob/development/contrib/drivers/ceph/ceph_metrics.go#L23
• A screenshot of an alert.
• Record demo with Dashboard

Orchestration and Automation

Xulin, Himanshu, and Ashit

Demo:
Bucket migration
Volume provisioning
Customized workflow, add script to do a step.
Dashboard integration (wait for dashboard code integration)

Object Lifecycle Management

Shufang and Neelam

Demo:
Manage object lifecycle with default tiers, using dashboard, migrating between AWS, Azure, Ceph, etc.
Show how a user define tiers? (Need to ask Shufang)
Recover API

File share

Pravin and Shruthi

Demo: Dashboard is ready to do demo

• Use dashboard to create profile, create a share, provide access, create a snapshot, using default NFS driver
• Manila Fileshare driver
• Provision volume for Kubernetes using CSI plugin

[xing-yang]

@xxwjj Please take a look of the demo steps. Thanks.