Hi all, I am in need of an infrastructure set up for data analytics / live video stream analytics application using big data and analytics technology..
The data is basically right now stored as structured data(no video streaming) in PostgresDatabase. ( Its an emergency call handling solution, In database which stores, caller info (address, mobile number, locations co-ordinates, emergency category metadata and dispatch information regarding rescue vehicles., Rescue vehicle location update (lat, long)every 30 seconds all are stored in the Postgres Database ..
Input1 : I have to do an analytics on these data( say 600 GB for the last 2 years its the size grown from initial setup).To perform an analytical application development( using python and data analytics libraries, and displaying the results and analytical predication through a dashboard application.)
Query 1. How much resource in terms of compute(GPU?(CPU) cores required for this analytical application? and memory ? And any specific type of storage(in memory like redis required ? ) etc, which I have to provision for these kind of application processing. ?? any hints most welcome.. Any more input required let me know I can provide if available.
In addition to the above I have to do video analytics from bodyworn cameras by police personnel, drone surveillance
Videos from any emergency sites, patrol vehicle (from a mobile tablet device over 5G )live streaming of indent locations for few minutes ( say 3 to 5 minutes live streaming for each incident. ) There are 50 drones, 500 Emergency rescue service vehicles, 300 body worn camera personnels.. and roughly 5000 incidents / emergency incidents per day happening, which needs video streaming for at least 1000 incidents for a time duration of 4 to 5 minutes live streaming.
Query2. What/(how many) kind of computing resources GPUs(CPUs)? RAM, Storage solutions I have to deploy in numbers( or cores of GPUs/how many/(CPUs)? RAM ? In Memory (Redis or similar ) or any other specific data storage mechanisms ? Any hints much appreciated..
10/30/2023, 6:49 PM
I’m not sure if this community will be able to help with that. There’s a ton of variables at play. For Postgres, specifically, you might want to join the Postgres Slack and ask there. But they’ll need to know a lot more about your query patterns, your DB schema, etc.
10/31/2023, 4:59 AM
Also you didn't exactly tell us what kind of analytics you want to do. There's a vast difference between different algorithms on how much compute capacity they need
Also, no mention of required computing time. You can use whatever and let it run for days basically...
But, in order for anyone to help (and like Paul said, this isn't the best place for that) we would have to know the specifics of the system and I'm sure that is it is a first responder emergency system then you simply can't tell that information. I'm actually thinking that by asking this at #google-cloud you have already told a bit too much of such a sensitive system.
10/31/2023, 2:14 PM
If any more inputs, I can provide . Sorry, If I am at a wrong place asked this.. in hope someone can shed some light ..
10/31/2023, 2:56 PM
Basically the best person to give you answers is the person who is going to do the analytics work. Ask him/her how much is needed and start with that