server power consumption, checks it against the power cap goal, and, if necessary, adjusts server performance to maintain an average power consumption that is less than or equal to the power cap goal. This functionality is available on all Intel-based ProLiant server blades.
Using the Insight Power Manager (IPM) v1.10 plug-in to Systems Insight Manager v5.1, customers can set power caps on groups of supported servers. The IPM software statically allocates the group power cap among the servers in the group. The group cap is allocated equitably among all servers in the group based on a calculation using the idle and maximum measured power consumption of each server. In addition, IPM can track and graph over time the actual power usage of groups of servers and enclosures. This provides data center facilities managers with measured power consumption for various time periods, reducing the need to install monitored PDUs to measure actual power usage in data centers.
Interconnect options and infrastructure
The BladeSystem enclosures easily enable connecting the ports of embedded devices to the interconnect bays. The c3000 Enclosure signal midplane (Figure 9) acts as a PCI Express (PCIe) bus connecting interconnect ports on blade devices to interconnect modules. It has eight device bay signal connectors (one for each half-height server blade and two for each full-height server blade) and four interconnect module connectors (one for each interconnect bay). The device connections are in groups of lanes. Each lane is a group of four pins (two sending traces and two receiving traces), resulting in full-duplex communication. This combination provides a 1x (500Mb/s) transfer rate with 2x = 2 lanes (1Gb/s).
Figure 9. Diagram of the HP BladeSystem c3000 signal midplane
By taking advantage of the similar four-wire differential transmit and receive mechanism, the signal midplane can support either network-semantic protocols (for example, Ethernet, Fibre Channel, and InfiniBand) or memory-semantic protocols (PCIe), using the same signal traces. Figure 10 illustrates how the physical lanes can be logically “overlaid” onto sets of four traces. Interfaces such as Gigabit Ethernet (1000-base-KX) or Fibre Channel need only a 1x lane, or a single set of 4 traces. Higher bandwidth interfaces, such as InfiniBand DDR, will need to use up to four lanes.
12
