A3Cube uses enhanced PCIe architecture to provide in-memory direct acceleration to existing applications, is compatible with the big storage OEMs rather than competitive with them. and goes to market entirely through channel partners.
A3CUBE, which emerged from stealth in February with RONNIEE Express, a new interconnect fabric which extends the capabilities of PCIe architecture, is now introducing it within an ultra-high performance hyper-converged system, Fortissimo Foundation.
Fortissimo Foundation is a clustered, pervasive, global I/O access storage system that linearly scales I/O, capacity and performance to provide an in-memory scale-out solution that aggregates all the resources in a cluster into a massive global name space. The trick is that it bridges all remote servers and resources to look and act as if they were local.
“The basic idea with what we are doing is very simple,” said Emilio Billi, A3Cube’s CTO and Founder. “There has been a much bigger interest in the last five years in accelerating data between the CPU and storage. Acceleration has been all local. When you need to scale out, that’s a problem. The new platforms like Hadoop for unstructured data are completely different from old structured databases because they are designed to be used in a clustered scale out environment.”
Fortissimo Foundation is designed to improve the performance of applications in unstructured data storage environments, to provide the power of high-end HPC clusters in everyday operations complete with the capability to access shared memory afforded by SMP systems minus the cost, complexity and limitations of building these scale out systems.
“Metadata synchronization on a single node creates bottlenecks, where the throughput is limited by a limited number of interfaces,” Billi said. “Fortissimo Foundation opens access to all resources in the cluster as if they were local, so the physical access time to the resources is not different if the resource is local or if it is remote. The result is that instead of a bottleneck limiting scalability in throughput, you have a point to point connection which solves this problem.”
Billi said this benefits all I/O intense applications.
“You can easily create in-memory Hadoop applications without buying extra software so you have a faster Hadoop solution with less cost,” he said. “You don’t need to buy HANA to perform analytics in-memory because you have up to 100% speedup with in-memory acceleration.”
The solution uses the RONNIEE platform with x86 compliant servers and any hardware the user wants. It creates server, disk and memory pools in a global namespace across all the resources, which is massively scalable, and which completely eliminates the metadata, so there is no single point of failure.
“The secret sauce is that we use enhanced PCIe with unique direct in-memory communications — RONNIEE Express – which provides direct memory access to local and remote resources,” Billi said. “It bypasses the OS kernel.”
Billi said that Fortissimo Foundation is positioned as a high end performance solution, very scalable, with high availability.
“We cover specific needs that are not yet covered by NetApp or EMC, so we don’t compete with one of their solutions,” he said. “This lets us work in a co-operative way with them. We are working with EMC, VMware, and with SIs that want to provide a unique solution to their customers.”
Billi said their go-to-market model is to work with SIs and resellers in a 100 per cent channel model.
“We are talking with six potential partners in North America now, including an integrator in Canada, and have more interest,” he said. “The idea is to start with about 10-12 partners in North America, then figure out how to scale out of North America. Europe is a maybe next year situation.”
Billi emphasized that right now, their solution is unique.
“We are the only one that provides in-memory direct acceleration to existing applications,” he said. “It’s plug and play, and your legacy application runs 10-20 times faster. It’s very unique and a key point of value.”