Want to try the Aprius Technology Evaluation Platform? Please contact us for more information on how you can test our various solutions.

Solutions based on Aprius technology maximize application performance and minimize infrastructure costs in a data center by addressing I/O bottlenecks at the server and the storage array. The company offers a simple approach to provisioning and managing I/O resources by enabling the PCI Express protocol to traverse over the Ethernet data center fabric.

Examples of PCIe-based I/O resources include:

• Flash storage cards
• Disk controllers
• Network interface add-on cards (NICs and CNAs)
• SAN interfaces (FC and iSCSI)
• Audio/video I/O cards
• Coprocessor/offload cards
• Utility I/O (e.g., telephony interfaces)

Aprius technology solutions simplify server deployments, increase I/O flexibility and dynamically allocate bandwidth and policies from pooled I/O resources accessible over the data center’s high bandwidth Ethernet network.  Aprius technology provides a number of key advantages to end-user customers:

By managing I/O resource configuration in software, such as network addresses and QoS settings, administrators can separate the logical management of resources from the physical hardware, simplifying datacenter management and significantly reducing the frequency of physical adds/moves/changes.

Improved application availability is achieved from the decoupling of server CPU/memory and I/O resources. As the components necessary to run a specific workload are physically decoupled, a failure of one component does not impact other components or prevent the workload from being executed.

Application deployment flexibility is increased via the ability to move I/O assignments with application workloads, even in VMM bypass configurations. Since any I/O resource, including non-virtualized cards such as hardware accelerators and offload devices, are effectively made mobile, they can be coupled to virtual workloads in a dynamic environment.

Hardware infrastructure economics are improved by higher utilization of fewer I/O resources (cards, cables and switch ports). I/O-related capital costs can be reduced by up to 50 percent in typical enterprise data center configurations, even if the customer is migrating from GbE to 10GbE infrastructure.

By utilizing a common transport fabric to reach high-value resources, the transport fabric is more effectively utilized. The benefits of a higher bandwidth, lower latency data center network can be realized with a faster return on investment. Especially in highly scaled environments, gaining efficiencies and removing costs from the edge returns a large overall cost savings.

The I/O resource pool also allows for n+1 redundancy to replace 2n redundancy, further improving the economics of consolidation. In this way, expensive I/O cards can be backed up by a single fail-over card accessible by multiple servers.

Removing physical I/O slots from server designs reduces excess power dissipation near critical and sensitive components (CPU/chipset/DRAM), while improving airflow to keep them cool. The result is higher compute density and extended lifetime and efficiency from core computing devices.

In typical enterprise data center environments, host I/O card-counts can add in excess of 30 watts per server. By removing just one card per server, rack power consumption can be decreased by 300 watts or more. Remote I/O allows even greater card consolidation, leading to a significant reduction in power and cooling costs across the data center.