For client engagements that include the deployment of hardware capacity for processing and/or storage, GraphStream uses COTS general-purpose servers as building blocks to deliver this capacity. We can deliver a wide variety of server configurations in response to specific client requirements. The majority of the servers that we are currently deploying are based on platforms from Intel and AMD that support the industry-standard x86-64 instruction set architecture.

The capabilities of the COTS servers that GraphStream provides are determined almost entirely by reference designs defined by industry standards and by the suppliers of the individual internal components, including:

• Server platform designs: Intel, AMD
• CPUs: Intel, AMD
• DRAM modules: Samsung, Hynix, Micron
• GPUs: NVIDIA, AMD
• Network interfaces: Mellanox, Intel
• Solid-state storage: Intel, Samsung, Micron
• Magnetic storage: Western Digital, Seagate

There are many competing suppliers of server products that are based on the platform reference designs listed above. Points of differentiation among these suppliers include chassis design; management and monitoring functionality; built-in self-diagnostic capabilities; support offerings; and price.

GraphStream's clients often require servers from a specific supplier, based on these points of differentiation and on their existing installed base. Hewlett Packard Enterprise is one of the suppliers whose server products are specified most frequently by GraphStream's clients. Each of these suppliers offers a wide selection of server models, each optimized for a specific range of use cases.

Hypervisor-based virtualization makes it possible to run multiple operating system instances on a single physical server. ScaleMP provides virtualization software that does the opposite, making it possible to run a single operating system instance across multiple physical servers, by aggregating the processing, memory, storage, and I/O resources of the physical servers into a single virtual server.