Building HTAP Applications with Python & Postgres—in Azure | PyCon 2020 Remote | Sai Srirampur

Demo of how to build HTAP applications on Azure—by scaling out Postgres with Hyperscale (Citus) on Azure Database for PostgreSQL. HTAP stands for “hybrid transactional analytical processing.” See how scaling out Postgres with Hyperscale (Citus) enables you to use a single database to serve both transactional and analytical workloads, reducing operational costs, and headaches, in a big way. Demo includes a lovely cameo by the Postgres extension to Azure Data Studio.

Transcript

1. Building HTAP Applications with
   Python & Postgres—in Azure
   Sai Srirampur | @saisrirampur @AzureDBPostgres @citusdata
   A Demo
   

   View Slide

2. Quick “fly by” about me / why am I giving this demo?
   Postgres team at
   Microsoft
   via Citus Data
   acquisition
   Creator of
   `django-multitenant`
   @saisrirampur
   

   View Slide

3. Let’s go / HTAP Demo on Azure / in 3 parts
   1 2 3
   @saisrirampur
   AZURE PORTAL HTAP APPLICATION AZURE DATA STUDIO
   

   View Slide

4. Azure Database for PostgreSQL / managed service
   @saisrirampur
   

   View Slide

5. APPLICATION
   SELECT
   FROM
   GROUP BY
   company_id,
   avg(spend) AS avg_campaign_spend
   compaigns
   company_id;
   METADATA
   COORDINATOR
   NODE
   WORKER NODES
   W1
   W2
   W3 …
   Wn
   SELECT company_id
   sum(spend),
   count(spend) …
   FROM
   campaigns_2009 …
   SELECT company_id
   sum(spend),
   count(spend) …
   FROM
   campaigns_2001 …
   SELECT company_id
   sum(spend),
   count(spend) …
   FROM
   campaigns_2017 …
   How Hyperscale (Citus) Shards Postgres Across Nodes / DISTRIBUTED AGGREGATE
   

   View Slide

6. Hybrid
   Transactional &
   Analytical
   Processing
   So what is an HTAP application anyway?
   @saisrirampur
   

   View Slide

7. Today’s Example of an
   HTAP application
   App that manages “Marketing/Ad
   Campaigns” for businesses
   • Create/Update/Delete Campaigns –
   Transactional Part
   • Analyze campaigns/clicks/impressions
   – Analytics part
   @saisrirampur
   

   View Slide

8. The challenge
   for HTAP
   applications
   @saisrirampur
   

   View Slide

9. Performance benefits of running your HTAP
   application on a Hyperscale (Citus) cluster on Azure
   Parallelism of scaled-out Postgres database
   cluster enables you to reduce costs & use
   single database for your HTAP workloads
   What I am going to show you today
   @saisrirampur
   

   View Slide

10. Let’s go / HTAP Demo on Azure / in 3 parts
    1 2 3
    @saisrirampur
    AZURE PORTAL HTAP APPLICATION AZURE DATA STUDIO
    

    View Slide

11. Most important takeaway from this Azure demo
    Citus open source mascot: Elicorn
    >part elephant, part unicorn
    >sharded, of course
    “
    https://aka.ms/hyperscale-citus-quickstart
    @saisrirampur
    

    View Slide

12. © Copyright Microsoft Corporation. All rights reserved.
    thank you
    merci
    grazie
    ευχαριστώ
    gracias
    ధన్యవాదాలు
    • Hyperscale (Citus) Quickstart Docs
    https://aka.ms/hyperscale-citus-quickstart
    • Multi-tenant Tutorial with Citus
    http://aka.ms/hyperscale-citus-multi-tenant-tutorial
    • django-multitenant library from Citus
    https://aka.ms/django-multitenant-github
    • Django sample app used in this demo
    https://aka.ms/django-PG-ad-app
    • Join Citus technical newsletter
    https://aka.ms/citus-newsletter
    Sai Srirampur
    @saisrirampur
    @AzureDBPostgres
    @citusdata​
    

    View Slide