Joviam utilises InfiniBand (IB) technology, unlike traditional cloud providers that use Ethernet. In laymen’s terms, InfiniBand is the 10-lane superhighway to Ethernet’s winding road, offering vastly enhanced latency, speed and throughout.
Commonly found in supercomputers (read: Large Hadron Collider), InfiniBand allows us to give our users significantly higher IO performance, all the while maintaining data consistency through synchronous replication. This is because IB enables hyperconvergence AND synchronous data replication on the one platform – Ethernet-based hyperconverged cloud platforms have difficulty performing synchronous data replication due to speed and throughput limitations.
Ethernet’s latency and lower throughput can bottleneck distributed systems on traditional cloud platforms. InfiniBand significantly reduces the latency of reading and writing data to other cluster servers through a combination of physical and logical architecture, and the use of RDMA for storage transport. For customers, full IOPS performance is unlocked with fewer committed resources and without costly ‘Provisioned IOPS’ agreements (see our official benchmarks results here).
Developers and those employing DevOps practices find these higher levels of compute performance enable faster production/delivery cycles and greater application performance with minimised cost.
InfiniBand (IB) is a high-speed network interconnect technology that effectively meshes nodes together for enhanced IaaS performance.
InfiniBand facilitates the connection and unification of servers, storage and networking via a single switch-based infrastructure into a centrally managed fabric. By unifying network interconnects, communication between storage and server is simplified, with processing, communications and storage managed as a single entity.
This combination of efficient fabric design and inherent RAS features (reliability, availability, serviceability), sees IB offer a superior IO capability complimentary to many cluster and grid environments – the processing power of many interconnected CPUs makes it favoured in high-performance and high-scale computing.
At Joviam, we aim to make this level of performance publically available to sizes of business and all types of use, not just ‘high-tech’.
We conducted official benchmarks utilising the industry accepted Phoronix Test Suite to measure the performance of our platform.
The results are below, and we think, speak for themselves.