Since we started in 2010, we’ve been listening to the market and our customers. Many still rely on relational databases, either on-premise, on the cloud, or as an MPP appliance. These databases haven’t developed as fast as the growth in data has.
SQream DB v2.6 lets you evolve with your data needs. Our fast GPU-accelerated JOIN lets you join on any column, without having to go through lengthy index creation procedures and manual tuning; SQream DB’s unique auto-compression feature means less worries about I/O; our high availability means even if you lose a node, you can still access your data; and SQream DB’s easy-to-use management UI and query editor lets you get up and running quickly.
Over the past few months, we’ve partnered with Alibaba Cloud to bring super-fast GPU databases that scale to one of the world’s largest cloud providers. We’ve also partnered with hardware and storage vendors, to put faster data analytics into the hands of more companies than ever. We’ve announced projects with some of the world’s largest telecoms for better customer interaction and improved network planning and monitoring.
And, from this week, SQream DB is available on AWS as a public AMI, with instance pricing starting at $0.9/hour.
New Features for SQream DB v2.6
- Incredible automatic GPU join optimization – speeds up some joins by up to 90%
- Better filtration and I/O reduction techniques can now filter calculated dates transparently
- We’ve slashed I/O and wait-times for queries that only return a small number of results, even when scanning trillions of rows
- Our unique auto-compression algorithms are improved with new schemes and better performance:
- Highly improved dictionary compression for text fields
- New compression schemes introduced for numeric fields
- We’ve added SQL functions for string operations – TRIM and REPLACE, to join the existing REGEX and string operations
- New date formats supported – Y/M/D and M/D/Y, on top of the standard ISO8601 date standards
Incredible join performance on the GPU
Since the first days of SQream DB, it was clear that no relational database can be successful without implementing the full set of JOIN operations that SQL users are accustomed to. SQream DB’s query engine which has been written from ground up to be columnar and run fully on the GPU. SQream DB runs aggregates, joins, filters, transforms, window functions, sorts all on the GPU using code written specifically for purpose. We consider this to be a very basic feature, but that shouldn’t be taken for granted! In fact, we are not content on joining just two tables like some other databases – we don’t limit the size or number of tables in the JOIN, nor do we limit the size of ORDER BYs or nested queries. And, we have been doing this since day 1.
One of SQream DB best abilities is therefore the ability to join any number of tables, on any number of keys, without having to pre-index or even match the data-types. And today, we’re happy to annouce that this feature is now better than ever, with up to 90% performance improvement, compared to before.
A natural, easy-to-use, GPU analytics database
We continue to build on SQream DB’s capabilities to provide a fast, flexible, and scalable GPU SQL database, with rich and standard SQL, as well as ACID compliance and high availability as core values. SQream DB v2.6 has the performance you’ve grown to expect from GPU databases, with the flexibility and scalability to adapt to your company’s natural growth.
SQream DB’s architecture is designed to work in conjunction with other databases, MPP appliances, BI tools, and machine learning systems. For example, SQream DB includes:
- A highly concurrent GPU architecture that is designed to handle large, heavy, and complex queries
- ODBC, JDBC, Node.JS, and Python connectivity for high-throughput connectivity to virtually any BI tool
- Full ACID compliance for serializability and durability, for you to build a reliable production environment for your business
We invite you to try SQream DB either on cloud, or on premise by visiting sqream.com/try-sqream-db