White Paper

As with Open19 Foundation, our strategies are based on a set of key infrastructure visions: • Unlimited bandwidth • Zero latency • Compute on demand • Disaggregation • Programmable data center • Self-healing While no technology has "unlimited bandwidth" or "zero latency," we want to enable network infrastructure performance where, for all intents and purposes, the applications (and users) effectively experience unlimited bandwidth and zero latency. Once achieved, on-demand compute resources can be made available anywhere. Every server on an ECMP [equal cost multipath routing] nonblocking fabric is essentially "equidistant" from every other server, and any server can potentially host any application. Application modules, functioning as Virtual Machines (VMs) and/or containers, are now free to run anywhere with equal access to the network, cooperating servers, and storage. The OCP and the Open19 Foundation have demonstrated this capability and enabled great flexibility for more than 250 participating entities, including major banks, internet, cloud, and telecommunications providers. According to Frank Frankovsky former Facebook, VP Hardware Design & Supply Chain: Bringing the hottest new CPUs into our environment can have a big impact on performance and efficiency. Why not just swap out the CPUs and leave the rest of the server and rack elements in place, rather than rolling in pre-packaged rack-loads of new servers?" By preinstalling fully cabled, bare racks to their data centers, LinkedIn technicians can install, provision, and bring online 96 servers in less than 90 minutes. Cloud providers like Equinix provide customers with the option to lease a single-server module as part of a bare metal cloud. 3 Shipping industry: The intermodal shipping container Prior to the ISO shipping container's introduction, products were manually handled in inefficient ways. The standardized ISO container solved these issues by abstracting the various products from the entire end-to-end shipping, storage, and transportation infrastructure. A Navy standard Hyper Infrastructure will define a set of standard, HyperComposable module configurations – micro containers – that provide this benefit in the datacenter in the same way the Intermodal Shipping Container does for transcontinental shipping. HyperComposable micro containers normalize size/volume, mechanical mounting, power, environmental resilience, and cooling. By abstracting physically encapsulated technologies from the infrastructure, standard modules can be easily removed and replaced by newer technology with zero associated shipboard industrial work. Older modules can be repurposed as spares or redeployed on other platforms that may not require or have a budget for the latest technologies. Auto manufacturing The current best practice in the auto industry is to develop a set of common extensible platforms that can be tailored to build specific vehicle types by adding modular assemblies: engines, bodies, drive trains, cockpits, etc. By the year 2020, 95 percent of ~33M units manufactured by the top 12 OEMs will be based on an average of three OEM core platforms each. Common platforms in the auto industry clearly have a proven track record of reducing cost and time to market. Similarly, a Navy-standard Hyper Infrastructure Common Modular Equipment Rack (CMER) will function as a universal platform for any number of applications. The strategy is to deploy CMERs at any number of end points of the platform HyperScale fabric. Applications developed in the lab on a specific combination of HyperComposable Modules can be deployed as the exact same set of HyperComposable modules, VMs, and/or Linux containers. An immutable binary file, in the form of a Linux container, can physically configure the CMSR module interconnect switch system. (See Figure 1: Disaggregation) Figure 1. Disaggregation: The trend is disaggregation of storage, servers and networks through open, modular, and software defined everything (SDx)

• Cover