📝 MY NOTES #009 - How One Online Bank Reduced Time to Insight and Transformed its Future
Theme : Hardware Transformation
Hello everyone,
Few weeks ago I had the chance to attend the Data Summit 2024 where speakers from different organizations spoke about a myriad of topics related to data. There were 10 sessions given by people who were CEOs, Data Managers, Data Engineers. Below are the titles of all the sessions I attended. In the coming weeks I will be posting my notes from each of these sessions & these titles will turn into links that you can use to jump between different session notes -
Data Insight in the Fast Lane: How One Online Bank Reduced Time to Insight and Transformed its Future
Building a winning data culture
Beyond the Basics: The last 10 things data teams think about
Building a shared foundation of trust through data storytelling
The dynamic duo of data and machine learning engineering in cybersecurity
The rise of the data generalist: Smaller teams, bigger impact
Navigating the maze of specialization in data engineering
Your Company's Success isn't Measured in megabytes, it's measured in impact!
Let’s start with the 3rd session I attended-
Data Insight in the Fast Lane: How One Online Bank Reduced Time to Insight and Transformed its Future by Deborah Leff - CRO at SQream
It feels great to share my notes on Deborah’s session, we recently connected over Linkedin. Deborah if you are reading this, great job explaining the Case study !
Deborah starts by introducing her background, after spending a decade at IBM working on Analytics & ML projects she moved to SQream.
Typical challenges of CPU based Environments
Because of the increasing capability of today’s AI & Ml models, Enterprises are seeing a downstream effect. The 4 main themes where these effects can be seen are -
Productivity : I have experienced this effect myself while working as a Data analyst where due to older systems and tools even a daily update of reporting would take 1.5 hour in SAS! This hugely impacts the amount of work that the data team can deliver at the end of the day.
Quality : Big point! I have worked on some POC projects where the amount of data we would use for building a model would be reduced given the amount of processing it would require to pull the data & on top of that the processing it would require to build a big model. It’s as if you are leaving a part of a movie out and then you expect an iMDB rating of 8+!
Innovation : Because of high costs related to computing many times the ROI of a project will never be achieved because the budgets will be cut shot given the huge ask for processing.
Cost : Deborah correctly puts it, if the current systems of an enterprise are overcharged, the cost of cloud architecture could be unbearable for companies.
The case of NCBA
A leading bank in Africa with over 60 million customers, growing at 20% a year.
The challenges
Infrastructural constraints - Legacy infrastructure which was incapable to withstand the increasing Analytics maturity in the company.
Lengthy batch processing - Refined data took 2 days to arrive because of current system’s latency leading to missed opportunities & badly affecting productivity.
Insufficient resource - Data team found difficult to keep up with the requests leading to build up of background hence increasing frustration within the team.
Key Objectives :
Remove bottlenecks
Improve productivity
Unleash innovation
Partnering with SQream & the results
This is where SQream comes into the picture, NCBA bank partnered with SQream and the following were the results of this collab -
The time it took the team to ingest updated data reduced by 80% which would mean that Analyst were able to play with the recent data from past day’s activity & hence add a lot of value in many activities of the business.
The time to Analyze which in the past suffered from latency issues saw almost 90% reduction in time to see Insights
Even with the improved performance the cost of licensing reduced by 73%! The CFO must have partied that day!
Deborah’s favorite! the Hardware footprint reduced by 97% from 66 nodes which is equivalent to 3 racks of data center down to just 2 GPUs!
How did the magic happen?
GPUs did!
Deborah proceeded to give a great introduction to the objective of GPUs -
GPUs over the years have become more powerful at processing but more importantly the cost of processing has gone down sharply in comparison to the processing power.
In terms of relationship of CPU & GPU, she adds
“The CPU is like the brain & the GPU is like the muscle”
SQream has created a proprietary software that combines these 2 important technologies, thus GPU & CPU work together to impact all elements of data processing from Ingestion to transformation to Analytics & GPUs can dramatically increase the speed of these processes.
SQream offererd the following to NCBA bank -
In this system SQream’s software (highlighted in green) acts as a turbo button in the existing system to make data processing faster.
SQream also offers the possibility to have complete Data Lakehouse solution which would greatly fuel innovation and data democratization within any organization.
Think of a data lakehouse like a big storage pond where you can keep all sorts of data. It's a mix of a data lake, which is a place to store raw data, and a data warehouse, which organizes data for analysis. So, in simple terms, a data lakehouse is a storage system where you can keep lots of different types of data and easily analyze them when you need to.
Very interesting case study! Looking ahead, as AI continues to evolve and play a more prominent role in businesses, it's evident that efficient data processing solutions like those offered by SQream will be essential for organizations to stay competitive and thrive in the data-driven future.
Are you currently exploring or considering GPU-based solutions to improve data processing speed and efficiency for your organization or even personally?
Finally, If you would like connect with Deborah, you can find her on Linkedin
Thanks for your attention & Hope you are enjoying this series on interesting webinars!
Until the next one!
Raghunandan 🎯
& If it’s your first time here, TheWeekendFreelancer currently has 5 ongoing series - Tools 🛠️, Maths 📈, Domain 🌐, Trends 📻 & My Notes 📝. Have fun reading!
P.S. - “The Weekend Freelancer” is a reader backed publication. Share this newsletter with your friends and relatives & consider becoming a free or paid member of this newsletter. Every subscription gives me an extra ounce of motivation to keep going 💪!
You can also support my work here. Cheers!