SQream
The only GPU data warehouse built
for any datasize and any workload
Business Development Consultant
Ung Hock Poh
Telephone
+66 (0)94-491-5442
ung.h@magicsoftware.co.th
TRANSFORMING FINANCIAL DATA ANALYTICS
Financial organizations require analytical solutions that provide significantly faster performance-at-sca
20X MORE DATA | 100x FASTER | 10% OF COST
Designed for terabytes to petabytes of data, SQream’s GPU-powered data warehouse enables financial organizations to rapidly ingest, join, and analyze data from varied sources, deriving insights based on up-to-the-minute data.
With unprecedented speed and flexibility, banks can uncover previously inaccessible intelligence about their business. Improve risk analysis, simplify reporting and increase customer loyalty – all by adding a cost-efficient and easy-to-use database to your existing workflow.
SQream can be deployed on-premise or on the cloud.
- Enhance fraud detection
- Refine risk analysis
- Optimize equity pricing
- Personalize client offerings

WHAT CAN YOU DO WITH 100X FASTER ANALYTICS?
Improve Compliance, Fraud Detection & Risk Management
New regulatory reforms require stringent reporting, often involving the calculation of complex metrics on raw data. SQream’s on-the-fly access to raw data, with multi-table JOIN capabilities on all felds makes calculations like XVA easy, even across billions of historical records. Similarly, SQream’s GPU-accelerated load-and-go design facilitates deeper analysis of account holders’ habits for more accurate risk assessments and tracking of historical trends.
SQream DB’s ability to correlate massive disparate data up to 100x faster than traditional systems allows fnancial organizations to identify abnormal activities or compliance issues before they cause damage.
Maximize Competitive Advantage
SQream enables organizations to dig deeper into historical data, performing queries on months of data instead of weeks, or years instead of months. Analytics windows can be expanded to build more accurate models of customer behavior, optimize pricing for financial products, and offer the right products to the right clients.
TAP INTO A WORLD OF NEW INSIGHTS

Fast Analysis of Massive Raw Data

Simple Deployment & Administration

Built for Your Growing Data

Cost-Effcient
COMPLEMENTING HADOOP FOR ANALYTICS
Hadoop is an increasingly important topic. Today’s solutions are either federating data in the BI layer or moving data into the data warehouse for mass access. New SQL engines on Hadoop are currently limited to performance boosts in federated scenarios. The real end game is to bridge both data stores with SQL. Over the next few years, this approach will become the mainstay of enterprises in deriving value from all their data.
SQream DB has an existing deployment scenario that works (side-by-side). To unlock the next deployment, some product development work is needed to integrate with the common industry formats, tools, and methodologies.

Key takeaways at a glance
- Hadoop is an infrastructure that has been widely implemented, despite being oversold.
- Data warehouses and DBMSs have failed to address the changing nature of data, and only allow querying of pre-computed, pre-aggregated data.
- Some companies, with a very wide range of data formats, benefit greatly from Hadoop.
- Some companies have been oversold on Hadoop, and have shoehorned it in, where a regular DBMS would have sufficed. This is where SQream DB will have the most value, as the business logic is still relational/structured for the most part. Hadoop will serve as another source of data for SQream DB.
SQream takeaways
- SQream DB has an existing deployment scenario that works (side-by-side)
- SQream DB has more potential that is currently untapped for hybrid deployment and unified deployment in the future. The product should be adapted to read data directly off Hadoop systems, with a reasonable effort (HDFS, federated queries to HBASE) to increase its maturity for the Hadoop ecosystem. This will allow it to be a strong contender for running queries with unified data layers in the future.
- We should conduct research to identify possible areas of improvement – how can SQream DB be used to read semi-structured data (arrays, nested objects) – like Snowflake and Vertica do already