A rules-based mechanism is described for powering down racks in an ordered and autonomous way in a data center. Power shelf controllers (PSCs), on different racks or on the same rack, communicate together through a network, called the PSC network, separate from the data network. The PSCs are aware of the other PSCs that share the same input power domain. When the racks are configured for use, each PSC is assigned a priority value, based upon the management provisioning layer assignment. Each PSC creates a table of all the other PSCs and tracks each assigned priority value. When a power event occurs, the PSC can power down components within the rack in accordance with the priority table. Recovery can also be carried out in conformance with the priority table.

A floating device location identification system includes a chassis defining floating device housings and including respective chassis location identification features adjacent each floating device housing that identify the relative location of that floating device housing. A floating device may be positioned in a first floating device housing and adjacent a first chassis location identification feature. The first floating device includes floating device cabling connector(s) that are connected via a cabling subsystem to a device location identification subsystem, and chassis engagement elements that are coupled to the floating device cabling connector(s) and that engage the first chassis location identification feature. The floating device transmits floating device location identifying information to the device location identification subsystem that is based on the engagement of the chassis engagement elements and the first chassis location identification feature, and that identifies a relative location of the first floating device housing in the chassis.

A server assembly service determines configurations for custom assembled servers based on time-series utilization metadata for servers executing workloads similar to workloads that are to be executed on the custom assembled servers. The server assembly service determines trends in the time-series utilization metadata and compares the identified trends to associations between workload utilization trends and application classes to determine one or more application classes for applications executing the workloads. The service uses the determined application classes to select server configurations for custom servers that are to be assembled to execute workloads similar to the workloads related to the server utilization metadata. In some embodiments, the service selects custom server configurations without access to applications or application data for workloads of concern. For example, the service may select custom server configurations without using profiling techniques that may intrude on customer privacy by requiring access to underlying applications or application data.

Technologies for connecting data cables in a data center are disclosed. In the illustrative embodiment, racks of the data center are grouped into different zones based on the distance from the racks in a given zone to a network switch. All of the racks in a given zone are connected to the network switch using data cables of the same length. In some embodiments, certain physical resources such as storage may be placed in racks that are in zones closer to the network switch and therefore use shorter data cables with lower latency. An orchestrator server may, in some embodiments, schedule workloads or create virtual servers based on the different zones and corresponding latency of different physical resources.

Embodiments are generally directed apparatuses, methods, techniques and so forth to select two or more processing units of the plurality of processing units to process a workload, and configure a circuit switch to link the two or more processing units to process the workload, the two or more processing units each linked to each other via paths of communication and the circuit switch.

A server rack with a plurality of compute nodes is positioned in a facility that includes a spine and the server rack includes a middle of rack (MOR) switch located near the middle of the server rack, vertically speaking. The MOR switch includes a plurality of ports that are connected via passive cables to the compute nodes provided in the server rack. In an embodiment the passive cables are configured to function at 56. Gbps using non-return to zero (NRZ) encoding and each cable may be about or less than 1.5 meters long. An electrical to optical panel (EOP) can be positioned adjacent a top of the server rack and the EOP includes connections to the MOR switch and to the spine, thus the EOP helps connect the MOR switch to the spine.

A server device is provided. The server device includes a substrate, a first server module, a first cable, a second server module and a second cable. The first server module includes a first server board and a first recognition unit, wherein the first recognition unit is connected to the first server board. The first cable is connected to the substrate, wherein the first cable is connected to the first recognition unit, and the first recognition unit generates a first recognition signal. The second server module includes a second server board and a second recognition unit. The second recognition unit is connected to the second server board. The second cable is connected to the substrate. The second cable is connected to the second recognition unit, and the second recognition unit generates a second recognition signal. The server device determines whether the server module is coupled to the correct joint.

Examples of the disclosure provide a smart angled mounting piece and an angled mounting piece assembly. By providing switching units and identification circuits in the smart angled mounting piece, when one or more of the switching units are triggered by the auxiliary member, an identification circuit electrically connected to the triggered switching unit and an identification circuit electrically connected to a switching unit that is not triggered collectively generate a logic signal characterizing the position at which the electronic device is installed to the cabinet, and the logic signal is transmitted to the electronic device through the communication module. The manager may know the position of the current electronic device in real time, without the need to specifically provide position acquisition elements on the cabinet, and without the need to install a separate management system on the back-end server, simplifying the difficulty of device management.

Technologies for allocating resources across data centers include a compute device to obtain resource utilization data indicative of a utilization of resources for a managed node to execute a workload. The compute device is also to determine whether a set of resources presently available to the managed node in a data center in which the compute device is located satisfies the resource utilization data. Additionally, the compute device is to allocate, in response to a determination that the set of resources presently available to the managed node does not satisfy the resource utilization data, a supplemental set of resources to the managed node. The supplemental set of resources are located in an off-premises data center that is different from the data center in which the compute device is located. Other embodiments are also described and claimed.

A datacenter network can be made of points of deliveries and patch panels. Rewiring the logical links within the datacenter network to meet a new network topology is computationally intense. Methods, systems, and apparatuses are provided to modify an existing network topology with a provided existing physical topology and logical topology into the new network topology. For example, the provided physical topology can include changes to the network, such as adding new points of delivery, adding additional patch panels, increasing the number of physical connections between points of delivery and patch panels, or removing a point of delivery.