Azure Marketplace Offering Gen 2 - ja-guzzle/guzzle_docs GitHub Wiki

Table of Contents

Overview

  1. Guzzle VM with pre-installed Guzzle that gets you up in a minute. Its fully self contained and uses:
  • Local storage for guzzle home
  • Local spark for compute
  • Local Guzzle repository - Postgres or mysql
  • Sample physical end point: Local file system as a source or target
  • Sample physical end point: Local postgres or mysql as a source or target
  1. Gets you up and running in few minutes without any addition input

VM init script

When VM is provisioned it should have init script to handle:

  1. Geneating self signed certificate in specified location which ngniix is reading (if we cant then lets pre-bundle it)
  2. Change index.html to point to FQDN of the VM

Inside Guzzle

  1. User will have UI flow to switch to Cloud storage instead of local - in this case existing Guzzle home (configs and log location copied over to cloud) and required guzzle binaries - primarily jars thirdparty and guzzle jars
  2. User will have UI flow to switch to external repo - No existing repo will be copied over from internal metastore to external
  3. Setup of external spark computes (user can't delete the local spark)- this is existing "Compute" screen. We may have to make it more interactive including able to test the spark connection , get the list of cost pools, DBR and VM options and
  4. Provide workflow for SSO configuration - and make it one way (once activated can revert)
  5. To do start/stop and other aspects of VM
  6. Using Nginix as push state server. Have nginix to bridge all the URLs to Nginix (https://github.com/ja-guzzle/docs/-/wikis/Running-Guzzle-using-nginx)
  7. Optimize the Atlas deployment if need be and make sure its seamless and works - lets make it must and not optional. Atlas should be installed and services are up. User can always stop/disable Atlas service through advance documentation.

Improvements (Future)

Below are areas which are beyond Marketplace offering - its more to improve the whole manageability around guzzle.

  1. Upgrade from UI
  2. Restart services from UI
  3. We can consider postgres instead of mysql
  4. Download the env configs which can be zipped for repo
  5. Key managements - at this point none as its all in physical end point
⚠️ **GitHub.com Fallback** ⚠️