SQream DB is an enterprise-ready solution for deploying analytics on Hadoop data. SQream DB provides data professionals a revolutionary way to explore and rapidly analyze their vast data.
- Reduced requirement for data preparation. SQream DB runs on any structured dataset in its raw form, reducing the need for indexing, pre-aggregations, etc.
- Nested GPU-accelerated JOINs with any number of tables and keys enable data consumers to correlate huge amounts of data.
- Dynamic workload management supports scheduled and ad-hoc analytics workloads from a variety of stakeholders, including dedicated performance for the most important users and processes.
- Fast loading and querying. SQream DB helps ensure high query concurrency while simultaneously loading new data, at up to 3.3TB/hour.
- Columnar engine and compression. SQream DB lets you gain significant advantage for I/O, performance, and storage by efficiently storing data in a columnar, compressed format.
- Auto-tuning database. SQream DB requires little to no maintenance. Most operations are completely automatic (with overrides for professionals), so data consumers can simply load and query data immediately.
SQream DB and Dell PowerEdge R940xa storage reference platform consists of the following components:
- One Dell EMC PowerEdge R940xa with 4 CPUs and 4 NVIDIA Tesla GPUs
- Expansions provided by compute-only Dell C4140 1U blades with 4 NVIDIA Tesla GPUs and NVLink
- ToR 10GigE ethernet switch for the BI fabric
- 100GigE or InfiniBand switches for the storage fabric for expansion
Figure 3. SQream DB on Dell PowerEdge R940xa reference architecture featuring additional PowerEdge C4140 compute nodes for scale-out