Dell PowerEdge VRTX Chassis Management Controller Version 1.0 for Dell PowerEd - Page 148
Power Supply Redundancy, Power Budgeting For Hardware Modules
View all Dell PowerEdge VRTX manuals
Add to My Manuals
Save this manual to your list of manuals |
Page 148 highlights
CAUTION: In AC redundancy mode, you must have balanced sets of PSUs (at least one PSU in each grid). If this condition is not met, AC redundancy is not possible. Power Supply Redundancy When power supply redundancy is enabled, a PSU in the chassis is kept as a spare, ensuring that the failure of any one PSU does not cause the servers or chassis to turn off. Power supply redundancy mode requires minimum two PSUs. Additional PSUs, if present, are utilized to improve power efficiency of the system if DPSE is enabled. Subsequent failures after loss of redundancy may cause the servers in the chassis to turn off. Power Budgeting For Hardware Modules CMC offers a power budgeting service that allows you to configure power budget, redundancy, and dynamic power for the chassis. The power management service enables optimization of power consumption and reallocation of power to different modules on the basis of demand. CMC maintains a power budget for the enclosure that reserves the necessary wattage for all installed servers and components. CMC allocates power to the CMC infrastructure and the servers in the chassis. CMC infrastructure consists of components in the chassis, such as fans, I/O module, and storage adapters, PCIe cards, physical disk, main board. The chassis may have up to four servers that communicate to the chassis through an iDRAC. For more information, see the iDRAC7 User's Guide at dell.com/support/manuals. iDRAC provides CMC with its power envelope requirements before powering up the server. The power envelope consists of the maximum and minimum power requirements necessary to keep the server operating. iDRAC's initial estimate is based on its initial understanding of components in the server. After operation commences and further components are discovered, iDRAC may increase or decrease its initial power requirements. When a server is turned on in an enclosure, the iDRAC software reestimates the power requirements and requests a subsequent change in the power envelope. CMC supplies the requested power to the server, and the allocated wattage is subtracted from the available budget. After the server is granted a power request, the server's iDRAC software continuously monitors the actual power consumption. On the basis of actual power requirements, the iDRAC power envelope may change over a period of time. iDRAC requests a power step up if the servers are fully using the allocated power. Under heavy load, the performance of the processors on the server may be degraded to ensure power consumption stays lower than the user-configured System Input Power Cap. The PowerEdge VRTX enclosure can supply enough power for peak performance of most server configurations, but many available server configurations do not consume the maximum power that the enclosure can supply. To help datacenters allocate power for their enclosures, the PowerEdge VRTX allows you to specify a System Input Power Cap to make sure that the overall chassis AC power draw stays within a given threshold point. CMC first makes sure that enough power is available to run the fans, I/O module, storage adapters, physical disk drive, main board, and CMC itself. This power allocation is called the Input Power Allocated to Chassis Infrastructure. After Chassis infrastructure, the servers in an enclosure are turned on. Any attempt to set a System Input Power Cap less than the "Power Burden" will not be successful. Power Burden is the sum of power allocated to the infrastructure and the minimum power allocated for the powered servers. NOTE: To use the Power Cap feature, you must have an Enterprise License. If necessary for the total power budget to stay below the value of the System Input Power Cap, CMC allocates servers a value less than their maximum requested power. Servers are allocated power based on their Server Priority setting, with higher priority servers getting maximum power, priority 2 servers getting power after priority 1 servers, and so on. Lower 148