by Angela Guess
A new press release reports, “Distributed analytics system startup Zeepabyte, Inc. announced groundbreaking benchmark results, including querying 3TB in less than three seconds, when testing its Cascade Analytics System on IBM Power Systems, on the Power Development Cloud. When compared to other market leaders, the test also demonstrated that Cascade performed twice as fast and used ten times fewer hardware resources, resulting in significant energy and cost savings. Data-intensive applications that leverage machine learning for artificial intelligence and advanced analytics require heavy computing power to deliver valuable insights in real time. To drive new efficiencies for customers working with large datasets, Zeepabyte created its Cascade Analytics System to provide distributed storage, distributed processing and real time query responses for big data, the Internet of Things (IoT) and business intelligence (BI) workloads. The company runs its Cascade Engine with IBM Power Systems on the Power Development Cloud to achieve peak performance for its clients, which include a major American automaker and East Coast utility companies.”
The release continues, “In its recent test, Zeepabyte used the Star Schema Benchmark, an adaptation of TPC-H benchmark for Data Warehouses, to measure query speed at large data scale and query filters of varied selectivity. Benchmark data sets of 1TB and 3TB were queried by Zeepabyte’s Cascade v2.0 engine at double the speed per core on IBM’s POWER8 cores when compared with Intel Xeon E5-2670 2.60Ghz cores. The total dissipated power of the IBM servers was less than half of that of Intel Xeon based servers. Cascade Engine has processed an order of magnitude more queries per core per hour than Microsoft’s SQL Server 2017 at 1TB database size and Actian’s Vector 5.0 at 3TB database size, both running on HPE ProLiant DL580 Gen9 at comparable total amount of processing power. In all benchmark tests compared to other TPC-H tests, Cascade was able to complete complex queries 4 and 2 times faster per core while using 20 and 14 times fewer CPU cores, respectively.”
Read more at PRweb.
Photo credit: Zeepabyte