SQreamDB Summer 2023 Release: Advancing Data Capabilities for the Data-driven Enterprise

By Ohad Shalev

10.2.2023 twitter linkedin facebook

We are excited to announce the latest releases of SQreamDB, the GPU-accelerated data analytics platform that enables fast and scalable analytics on massive datasets. In this release, we have added several new features and enhancements that make SQreamDB even more powerful and easy to use. Here are some of the highlights:


  • Acceleration by H100: SQreamDB now supports the NVIDIA H100 GPU, a high-performance computing card. The H100 GPU has 32GB RAM of HBM2 memory and 5120 CUDA cores, making it ideal for data-intensive workloads that require high throughput and low latency. With SQreamDB, you can harness the H100 GPU to run complex queries and massive ML and AI workloads faster than ever before. As an example, check out our 3X performance improvement running the TPC-H benchmark with a scale factor of 2500 – 

  • ARRAY data type: SQreamDB now supports a variety of functions with the ARRAY data type that allow you to manipulate and query arrays of values in SQL. Now you can create, access, modify, aggregate, and search ARRAYs, as well as perform operations such as concatenation, slicing, sorting, filtering, and more. Supporting the ARRAY data type enables you to simplify your queries and perform complex analytics on semi-structured data. You can also use ARRAYs as a preferable data source for your ML pipeline to store and process vector or tensor data for AI and ML applications.
  • SQLoader is now on GA: SQream’s CLI tool for loading data from other databases is now generally available. SQLoader simplifies the data migration process and allows users to customize the type of loading into SQreamDB – full table loading, CDC (change data capture), or incremental loading. SQLoader will first support ingesting data from Oracle and PostgreSQL databases, with more databases to come in the future. 
  • Metadata server optimized performance: SQreamDB’s metadata server based on RocksDB is responsible for managing the metadata of all the tables and partitions in the cluster. In this release, we have optimized the performance of the metadata server by allowing users to configure and tune it while reducing its memory footprint and improving its concurrency handling. This results in faster metadata operations such as creating or dropping tables or partitions, as well as better scalability and stability of the cluster.


  • GDPR-compliance: SQream is now fully compliant with the General Data Protection Regulation (GDPR), the European Union’s data privacy law. This means that we respect the rights of our customers and users to control their personal data, and we have implemented the necessary measures to ensure that our data processing is lawful, fair, and transparent. We have also updated our privacy policy and terms of service to reflect these changes.
  • Role-based Access Control (RBAC): RBAC allows you to grant or revoke permissions to users or groups based on their roles and responsibilities. You can use RBAC to control access on a granular level for specific objects – database, schema, table, function, view, column, or service – so you can ensure that only authorized users can access or modify the data they need. RBAC also helps you comply with data governance and privacy regulations by enforcing policies and auditing actions on your data.


  • Oracle Marketplace: SQream is now available on the Oracle Cloud Marketplace. With SQream’s high-performance analytics engine integrated into Oracle Cloud, you can turbocharge your data analytics, extracting insights faster and with unmatched scalability.
  • New deployment at Samsung Cloud Platform (SCP): We proudly announce that SQreamDB is now available on the Samsung Cloud Platform, a leading cloud service provider in Korea. Samsung Cloud Platform offers a comprehensive suite of cloud solutions, including compute, storage, network, security, database, and AI. By deploying SQreamDB on the Samsung Cloud Platform, you can enjoy the benefits of a service that scales with your needs and provides high availability, reliability, and security. You can also take advantage of the SCP’s AI capabilities to integrate SQreamDB with other AI services and tools.
  • RHEL8 and CUDA 11: SQreamDB can now be deployed on RedHat Linux 8 servers and supports Nvidia’s CUDA 11 toolkit.



The SQreamDB Summer 2023 Release is here and it’s all about you. Experience improved load times that further power your data-driven decisions, making every insight a bit more impactful. With updated compliance tools, your business can better traverse the global regulatory landscape. Our refined deployment capabilities allow you to integrate technology smoothly into your operations, easing the journey into a more data-oriented organization. SQreamDB isn’t just a tool—it’s your companion in turning data points into opportunities, reshaping the future one query at a time.

To learn more about SQreamDB and how it can help you accelerate your analytics and ML projects, please visit our documentation.