A3Cube Announces Exa-Converged Parallel Systems

Print Friendly, PDF & Email

hadoop_F4000Today Italy’s A3Cube announced the F-730 Family of EXA-Converged parallel systems built on Dell servers and achieving sub-microsecond latency through bare metal data access.

A3Cube’s EXA-Converged infrastructure represents the next step in the evolution of converged systems”, said Emilio Billi, A3Cube’s CTO, “while keeping and improving on the scalability and resilience of Hyper-Converged infrastructure. It is engineered to converge all system resources and provide parallel data access and inter node communication at the bare metal level, eliminating the need for, and the limits of, traditional Hyper-converged systems. The system can efficiently use all the fastest storage devices currently on the market or planned to come to market, and puts all existing solutions in the rear view mirror.”

The F-730 Family is the first generation of EXA-Converged solutions specifically designed for Big Data, Analytics and IO intense workloads. It is based on a unique and revolutionary architecture engineered to efficiently address the computational and storage needs of current and future applications where time to result is a crucial and indispensable requirement. F-730 eliminates the need for expensive and complicated external storage and ensures the fastest possible data access through bare metal convergence of memory, storage, and IO, as well as computation and inter-node communication that benefit from remote direct memory access (RDMA).

Other solutions currently available on the market are known as Hyper-Converged and use a Hypervisor to virtualize all hardware resources. This means an additional software layer that doesn’t permit direct access to the hardware resources and results in an unavoidable slowdown of performance (extremely high latency, milliseconds, in data access and undesired continuous CPU content switching) and the inability to take full advantage of new storage technologies such as SSD, NVMe and 3D XPoint technology co-developed by Intel and Micron.

a3cubeA3Cube’s F-730 Family is a fully clustered parallel solution where multiple F-730 nodes are efficiently combined to create powerful systems that scale linearly in all aspects of performance: data IO and data capacity. The F-730 Family employs A3Cube’s proprietary HSDS architecture (Hardware accelerated Software Defined System), in combination with A3Cube’s Direct Pervasive IO architecture, to remove the existing bottleneck between all the resources distributed across the clustered nodes. The F-730 Family integrates FortissimoFSTM Engine, a high-performance clustered low-latency manager that aggregates the system resources at bare metal (using a proprietary technology that is called direct pervasive IO) and provides optimized convergence of memories, storage, computation and RDMA.

The FortissimoFS Engine guarantees ultra-fast data access to many modern applications including Hadoop, Spark and other analytics platforms and frameworks through the bare metal replacement of HDFS which results in an overall boost in performance ranging from 6X to more than 10X compared to the standard HDFS implementation ( benchmarks available such as TeraSort).

Machine learning, drug discovery, DNA-related studies and large-scale data mining are among the many Big Data and Analytics applications that will benefit greatly from performance improvements via the F-730 Family.

The F-730 Family includes:

  • F-730R based on PCIe RONNIEE express “in-memory” network
  • F-730I based on Infiniband interconnection
  • F-730 TurboIO, database accelerator for MySQL, MariaDB and Oracle DB. The solutions can scale from two nodes to more than 1,000 nodes and from a few terabytes to multiple petabytes of converged storage.

Sign up for our insideHPC Newletter