What are you looking for ?
Infinidat
Articles_top

A3Cube Data Plane Encapsulated in NIC: Ronniee Express

Transforming storage networking to eliminate I/O performance gap between CPU power and data access performance for HPC, big data and data center

A3Cube Inc., a developer of architectures for exascale high-performance storage systems, announced a ‘brain inspired’ data plane encapsulated in a Network Interface Card (NIC) designed to bridge supercomputing benefits to the enterprise by transforming storage networking to eliminate the I/O performance gap between CPU power and data access performance for HPC, big data and data center applications.  

 Designed for performance
A3cube,RONNIEE Express

The RONNIEE Express data plane elevates PCIe from a simple interconnect to a new intelligent network fabric, leveraging the ubiquity and standardization of PCIe while solving its inherent performance bottlenecks.

A3Cube’s In-Memory Network technology allows direct shared non-coherent global memory across the entire network, enabling global communication based on shared memory segments and direct load/store operations between the nodes. The result is low possible latency, scalability and disruptive performance that is orders of magnitude beyond the capabilities of today’s network technologies including, Ethernet, IB and FC.

Organizations struggle to keep up with the amount of traffic on traditional networks generated from a variety of sources,” said Bob Laliberte, senior analyst, ESG. “A3Cube’s In-memory Network fabric leverages an innovative approach to transforming HPC, big data and data center environments in order to drive greater performance and efficiencies in the network and storage systems. A3Cube is extending PCIe capabilities in order to deliver a next generation network that it claims will overcome traditional network bottlenecks utilizing a high performance (Nano-second latency) and massively scalable architecture.”

The data plane enables exascale storage that combines supercomputing’s massively parallel operational concepts and an I/O interface eliminating central switching, thanks to the support of a multi-dimensional topology like 2D/3D Torus and Hypercubes. This reduces network overhead, slashing the latency of traditional storage networking designs and introduces military grade reliability along with carrier grade data plane features.

The RONNIEE Express communication mechanism creates a paradigm shift in network communication that introduces a full application with a transparent memory-to-memory direct connection. The In-Memory Network discards the protocol stack bottleneck and replaces it with a direct memory-to-memory mapped socket, producing disruptive performance enhancements while leveraging commodity hardware.

Today’s data center architectures were never designed to handle the extreme I/O and data access demands of HPC, Hadoop and other big data applications,” said Emilio Billi, founder and CTO, A3Cube.  “The scalability and performance limitations inherent in current network designs are too severe to be rescued by incremental enhancements. The only way to accommodate the next generation of high performance data applications is with a radical new design that delivers disruptive performance gains to eradicate the network bottlenecks and unlock true application potential.

A3Cube’s first three products incorporating the RONNIEE Express product line address different data center requirements in building out an In-Memory Network fabric and include:

  • RONNIEE 2S is a compact PCIe-based intelligent NIC designed to maximize application performance using a combination of hardware and software. It eliminates conventional communications bottlenecks and provides multiple channels with <1µ and fast direct remote I/O connections with nanoseconds level latency.
  • RONNIEE RIO is the first general purpose NIC supporting Ethernet and memory-to-memory transactions in a 3D torus topology that can plug in any server equipped with a PCIe slot. This data fabric is designed to deliver performances and presents a scalable interconnection fabric based on a patent pending shared memory architecture that implements the concept of distributed non-transparent bridging to extend PCIe features and benefits over a next generation network architecture.
  • RONNIEE 3 is a card that is designed to extend the scalability of RONNIEE 2S and optimized for high performance data environments. The In Memory Network provides support for memory-to-memory transactions without the usual software overhead to achieve efficiency and performance compared to ordinary interconnection fabrics available.

A3Cube was founded in 2012 as a result of more than five years of R&D. The company has assembled a team of engineers in hardware, firmware, software and system design engineering and management with track records in super computing and HPC environments.

All design and development activities are internal to the company in order to maintain control of all aspects of the company’s products.

Comments

Comment by Stacy Kenworthy, former founder and CEO of ApplianceWare:

They are solving the network side of the problem. The challenge they will face is that Intel has provided an alternative solution for the network interface that uses the host CPU. 

More on the subject

With this technology, essentially all 10Gb/s NICs become obsolete since Intel can handle the network interface layer with the existing CPU thus removing the need to accelerate this portion.  Maybe there is a play at 100Gb/s which is coming shortly .

The bottleneck is not at the network layer but at the file system layer. From a technical standpoint there are just too many moves and operations in memory requiring too many clock cycles to handle file system operations. This is before things land on disk where all the operations have been completed. The problem is that an OS will NEVER (in my opinion) be able to keep up with 10Gb/s speeds as it bogs down the CPU too much. You can fan this out across 5 or 10 CPUs but then you are fragmenting the data and create additional overhead in managing and coordinating the streams to disk.

 

Articles_bottom
AIC
ATTO
OPEN-E