When deploying racks, rackable servers, tower servers, direct attached storage or other devices, the user needs reference parameters to calculate power consumption and electric current values for different configurations. Also, sales and IT administrators need a tool to understand detailed device information.
The main purpose of the Capacity Planner tool is to provide data maintenance, power consumption and electric current calculations, as well as summary reports, to help IT administrators manage racks and devices. The information and reports will significantly increase deployment efficiency and also help sales people conveniently introduce products to customers. The pictures in this document show page layouts and displays. The data in the pictures are only for reference and may differ from the actual data.

User Help

1 Option Settings

This tool provides an interface to configure some settings upon startup. Language settings need to be configured on the option settings page.

option

After a language is selected, the user interface will switch to the selected language.

2 Page Layout

There are five areas on the page: 1)Deployment area, 2)Toolbox, 3)Summary area, 4)Power area and 5)List area.

  • The Deployment area is used to operate racks and devices.
  • The Toolbox offers five icons for different functional operations.
  • The Summary area is used to display configuration and power summaries.
  • The List area provides five lists: rack list, rackable server list, tower server list, density server list, direct attached storage list and other device list.
  • The Power area is used to display general power information for racks and devices in the Deployment area.
layout

3 Add Rack

You can select a rack from the rack list and drag it to the deployment area. This operation can be also accomplished by double-clicking or right-clicking a rack and then selecting "Add" in the popup menu. A rack without any deployed devices will appear in the deployment area after the "Add" operation.

addRack

The detailed configuration and power information for the selected rack will be shown in the summary area.

4 Rack Operation

You can operate on a rack in two ways: Right-click the rack in the deployment area to display an operation menu, or click the icon buttons on the upper-right of the rack.

rackOperation

You can perform three rack operations: Copy, Configure and Delete.

  • Copy: Copies the selected rack containing all existing devices. If you create too many copies, the performance of the tool will be affected. Due to browser limitations, too much device information cannot be saved. Please pay attention to the number of racks.
  • Configure: Opens the Rack Configuration page to modify configuration settings.
  • Delete: Deletes the selected rack including all devices in it.

Select the Configure item in the menu or click the Configure button when a rack is selected to display the rack configuration dialog box, as shown in the following figure:

deviceConfig

Here, you can configure the Rack Name, Max Current, Max Weight and Rack Voltage. If a server is deployed in the rack, its voltage will be the same as the rack's voltage and cannot be changed. If a server is deployed as a standalone server in the deployment area, its voltage can be changed.

5 Add Device to a Rack

You can add devices to a rack using the following procedures:

  • Open a device list (rackable server list, density server list, direct attached storage list or other device list) and select and drag a device to the deployment area.
  • Double-click a device.
  • Right-click a device and select "Add" in the popup menu.
addDevice

When selecting a device in a rack, the summary area will display the configuration and power summary of the selected device.

When adding a user-defined device to a rack, a pop-up dialog will be shown to prompt you to input the configuration information.

deviceConfig

6 Device Operation

You can operate on devices in two ways:

  • Select a device in a rack and right-click. This displays the operation menu.
  • Select a device in a rack and then click one of the icon buttons to the upper-right of the rack.

You can perform three device operations: Copy, Configure and Delete.

  • Copy: Copies the selected device in the rack.
  • Configure: Configures the selected device. For the detailed rackable server configuration information, please refer to the instructions for configuring a rackable server.
  • Delete: Deletes the selected device.

7 Rackable Server Configuration

Items that Influence Server Power Consumption

influenceConsumption

There are many aspects of a server that can influence the AC power consumption. When a Thinksystem server is being configured in Lenovo Capacity Planner (LCP), these points should be considered:

  • Hardware configuration: Not all hardware is created equal. For example, different CPU SKUs have different TDP wattages. Different memory DIMM capacities consume different amounts of power. But, even beyond the high-level choices for things like CPU SKUs, DIMM size, network card, etc, there are finer points to the hardware configuration that can influence the AC power consumption. Some examples include:

    • Lower speed DIMMs consume lower power.
    • Generally speaking, for a given total system memory size, populating 2 DIMMs per channel will lower the total memory power. Effectively, the number of DIMM channels are reduced in half. Less memory channels operating means lower power. But it also can reduce memory performance, depending on the workload.
    • Power consumption among networks adapters of the same type, but different vendors, can vary widely. Unless a specific vendor is required, it’s best to create LCP configurations with different choices of adapters and compare the total power draw.
    • In general, for a given size total size of storage, solid state storage will draw lower power than traditional HDDs with spinning platters. Solid state storage contains no motor which is required in traditional HDDs to spin the platters.
    • When possible, pick a power supply with 80+ Titanium efficiency, to minimize the AC power consumption.
  • Power supplies (PSUs) have bell shaped curves where the efficiency peaks near the 50% load point and tapers off below or above 50%. In addition, a server’s DC load gets split equally among all the operational PSUs. Armed with this information, for a given hardware configuration and workload level, it’s possible to select a PSU wattage rating such that each PSU is operating near it’s peak efficiency.

    For an N+N power policy, take the maximum DC power draw for the server and divide it by the number of installed PSUs. Then, pick a PSU wattage rating close to 2x that number. For example:

    • If a server’s maximum DC power =2000W and N+N with 4 power supplies is desired, then:
    • Optimal PSU =2*(2000W DC maximum power) / (4 PSUs) =1000W
    • Pick a PSU rating close to 1000W, but at least 1000W (e.g. Lenovo 1100W PSU)

    For an N+1 power policy, take the server’s maximum DC power and divide it by the number of PSUs. Then take the server’s maximum DC power and divide it by N. Whichever result is larger is the target for the PSU rating. For example:

    • If a server’s maximum DC power =2000W and N+1 with 4 power supplies is desired, in this case N=3, and:
    • 2000W/(4 PSUs) =500W
    • 2000W/N =2000W/3 =667W
    • 667W is the larger of the two numbers. Pick a PSU rating close to 667W, but at least 667W (e.g. Lenovo 750W PSU)

    For a N+0 (non-redundant) power policy, select a PSU rating close to the server’s maximum DC power divided by the number of installed PSUs. For example:

    • If a server’s maximum DC power =2000W and N+0 with 4 power supplies is desired, then:
    • Optimal PSU =(2000W DC maximum power) / (4 PSUs) =500W
    • Pick a PSU rating close to 500W, but at least 500W (e.g. Lenovo 550W PSU)
  • In LCP, select the maximum AC operating voltage that the datacenter can tolerate. As the AC voltage goes up, power supply conversion efficiency increases.
  • If the datacenter ambient temperature can be to run at approximately 25C or less, do not check “Fan Max Speed” in the server configuration page of LCP. As fan speed increases linearly, the power draw increases exponentially. If a cooler datacenter environment can be maintained, substantial fan power can be saved, depending on the server selected. Of course, this it not applicable if Lenovo’s highly efficient water cooled servers with no fans are used.
  • Where possible, use liquid cooled servers: Direct liquid cooling is significantly more efficient compared to blowing air across heatsinks and electrical components. Also, any additional cost to implement liquid cooling in a datacenter can typically be offset by the electrical cost savings within a few years. The most efficient servers have complete liquid cooling with no fans (e.g. Lenovo DW612 Enclosure + nodes). Second best are servers that use a combination of liquid and air cooling (e.g. Lenovo SR675 V3).
  • Turbo mode: When turbo mode is engaged on a CPU, it allows the CPU to opportunistically run faster than its rated frequency. However, when turbo mode is initially engaged, the CPU power can spike to 120% of its rated power for up to 10 seconds. And as the turbo frequency increases linearly, the power increases exponentially. Therefore, turbo mode is less efficient than non-turbo mode. If there is a desire to minimize power consumption or maximize efficiency, disable turbo mode. In Lenovo Capacity Planner, this is done by setting the Time Duration to Long Term in the server configuration page.
  • Workload level: In LCP, set the Load Factor to closely match the actual workload that will be run on the server. Most servers rarely run above 70% for a long period of time.
  • UEFI and operating system settings: The idle power estimate in Lenovo Capacity Planner assumes that all power management features are enabled in the server UEFI settings and the operating system settings. Additional power savings may be realized if it is acceptable to the customer to enable functions for which there is a known errata. Lenovo does not enable those features by default because, depending on the errata and workload, it can affect the stability of the server.
  • End-to-end conversion efficiency for each server component (e.g. CPUs): Lenovo Capacity Planner uses realistic conversion efficiencies for the installed power supplies (PSUs) and voltage regulators (VRDs) in the server. In addition, LCP also factors in the workload level when calculating the conversion efficiency. It is possible to lower a server’s AC power estimate if it is assumed that the PSU and VRDs are always operating at or beyond their peak efficiency. This is particularly true for the CPUs, which typically draw the highest percentage of power for a server. Lenovo, however, does not condone and does not use this approach, as it will generate unrealistic power estimates and lead to datacenter decisions being made based on inaccurate data.

You can modify the rackable server's Server Name, CPU, Memory, Storage and other components.

  • The rackable server component restrictions and thresholds should be considered during configuration.
  • Click "OK" to save the configuration settings.
  • Click "Cancel" to cancel the operation.
  • Click the "Tip" button to display prompt information.
serverConfig

About the Term Duration, you can refer to below table:

Power Level Conditions Description Text
Idle long term time (but really doesn’t matter) +idle fans All power management features are enabled in UEFI and the operating system. Server is booted to operating system and sitting idle for several minutes.
Load Factor long term time +nominal fan with user selected load factor (e.g. 70% default) Typical long term power consumption under nominal conditions with no faults and user selected load factor.
Nominal Max long term time +nominal fan power with 100% load Typical long term power consumption under nominal conditions with no faults and 100% load.
Worst Case Max short term time +max fan power with 100% load Maximum power consumption used for power supply sizing. Typically encountered under fault and/or high temperature conditions.
  • Long Term Time Duration represents the max power level of the system over several minutes for the load factor selected.
  • Short Term Time Duration reflects the power level that the server can get to for the first 10-20 seconds after a workload is initially started.

The Load Factor slider is used to configure the expected power utilization. The Power Utilization can show you the PSU’s capacity based on the different server configurations and PSU’s policies. When you configure a server on the Server Configuration Page, if a type of component cannot be supported or it cannot be deployed with other types of the component on the server, it will not be displayed in the component selection lists on the left of the page, such as CPU, Power Supply etc. If it is necessary to add the desired component, modify the configuration so that the desired component can be added. 

Click the "CPU benchmark" button to display the CPU benchmark page.

  • Click CPU2017 to display the corresponding CPU benchmark data in CPU2017. CPU2006 is the same.
  • In the CPU selection box, you can search for a CPU to view its benchmark data.
  • Click the button at the top right of the page to open the help page.
  • Click the OK to close the CPU benchmark page.
  • The hint information* shows the source of the CPU benchmark data.
serverConfig

About balanced memory:

Click the "Capacity checking" button to display the capacity page.

  • You can check the available power of the currently selected PSU or PSUs.
  • If Available Power is displayed as "N/A", this indicates that redundancy mode is not supported by the currently selected PSU or PSUs
  • Selecting one item and clicking 'OK' will update the current power supply configuration.
  • Click the "Cancel" button to close the capacity page without saving.
serverConfig

Click the "Add User Defined Component" button to display the defined component page.

  • You can define the Name, Idle power, and Stress power of the component to be added.
  • In the Adapter Cards Type selection box, you can select the type of definition component to be added.
  • Click the OK to add the defined component.
  • Click the Cancel button to close the defined component page without adding.
serverConfig

8 Add Rackable Server

If there is no rack in the deployment area, select a rackable server from the rackable server list and drag it to deployment area; or double-click a rackable server; or right-click a rackable server and select "Add" in the popup menu. The rackable server will be displayed in the deployment area immediately.

AddServerAlone

There are two ways to invoke rackable server operations: right-click the rackable server in deployment area to display the operation menu; or click the icon to the upper-right of the selected rackable server.

You can perform two rackable server operations: Configure and Delete.

  • Configure: Configures the rackable server in deployment area.
  • Delete: Deletes the rackable server in the deployment area.

9 Density Server and Flex System Configuration

After a density server or flex system is added, select a server to be added to the density server or flex system from the server list and drag it to deployment area; or double-click the server; or right-click the server and select "Add" in the popup menu. The selected server will be added to the density server or flex system.

    serverConfig

When the density server or flex system is selected, click the "Configure" button to modify the density server's or flex system's Server Name and power settings. When you configure a server on the Server Configuration Page, if a component type cannot be supported or it cannot be deployed with other component types on the server, it will not be displayed in the component selection lists on the left of the page, such as Power Supply. If it is necessary to add the desired component, modify the configuration so that the desired component can be added. 

  • The density server and flex system component restrictions and thresholds should be considered during configuration.
  • Click "OK" to save the configuration settings.
  • Click "Cancel" to cancel the operation.
  • serverConfig

Select the server in the density server or flex System using the right-click menu or button to the right.

  • You can modify the server's Server Name, CPU, Memory, Storage and other components. When you configure a server on the Server Configuration Page, if a component type cannot be supported or it cannot be deployed with other component types on the server, it will not be displayed in the component selection lists on the left of the page, such as CPU. If it is necessary to add the desired component, modify the configuration so that the desired component can be added. 
  • The servers components restrictions and thresholds should be considered during configuration.
  • Click "OK" to save the configuration settings.
  • Click "Cancel" to cancel the operation.
  • serverConfig

10 Create a New Configuration

Click the "New Configuration" icon in the toolbox in the top-right corner of the page to create a new configuration. If there is any existing configuration in the deployment area, its information will be deleted by this operation.

When creating a new configuration, you will see the following pop-up dialog box:

create

There are three buttons in the pop-up dialog box: Save, Continue and Cancel.

  • Click the “Save” button to display the save popup window.
  • Click the "Continue" button to discard the current configuration.
  • Click the "Cancel" to cancel this operation.

If this tool runs in offline mode and is opened in IE, the Save button cannot be used.

11 Open Configuration
open

You can click the "Open or Import" icon in the toolbox in the top-right corner of the page. Click "Open configuration..." to view and load configuration files. When loading a file, the current configuration will not be automatically saved. If you want to save the current configuration, please refer to the instructions for saving a configuration file.

open

Select the configuration file you want to load from the file list, and click "OK" button to load the selected file.
Select a file in the file list and click "Delete File" to delete it.
This operation is not supported in offline mode using IE.

12 Import Configuration

You can click the "Open or Import" icon in the toolbox in the top-right corner of the page. Then, click "Import configuration...".

open

When loading a file, the current configuration will not be automatically saved. If you want to save the current configuration, please refer to the instructions for saving a configuration file.

13 Import CFXML

You can click the "Open or Import" icon in the toolbox in the top-right corner of the page. Then, click "Import CFXML...".

open

When loading a file, the current configuration will not be automatically saved. If you want to save the current configuration, please refer to the instructions for saving a configuration file.

14 Save Configuration
open

You can click the "Save or Export" icon in the toolbox in the top-right corner of the page. Click "Save configuration..." in the toolbox in the top-right corner of the page to save the current configuration in the deployment area.

Enter the file name in the popup dialog box. A warning prompt will be displayed if the entered file name is already in use.

save

Enter file name, and click "OK" to save current configuration file.
This operation is not supported in offline mode using IE.

15 Export Configuration

You can click the "Save or Export" icon in toolbox in the top-right corner of the page. Then click "Export configuration..." button.

open

16 Report

Click "Create Report" button in the toolbox in the top-right corner of the page to view the current configuration and print out the configuration report.

The popup dialog box displays all configuration information for the deployment area.

report

17 Magnification

Devices in a rack can be shown in Magnification view.

If you move the mouse over a device (rackable server or user-defined device) in a rack and hover over it for several seconds, the magnification view will be enabled.

magnifying

Move the mouse off of the device to disable magnification view.

18 Total Cost of Power Settings

Total Cost of Power Settings:

  • Cost per kWh: The Cost per kWh.
  • PUE (Power Usage Effectiveness): The efficiency of power usage.
  • Server Lifecycle (Years): The lifecycle of the server.
  • PCF Emission Factor (kg/kWh): The product carbon footprint emission factor.
magnifying

19 Power Summary
Field Definition
System Input Power (W) AC or DC input power.  Equal to the Server DC Power Consumption (W) divided by the PSU efficiency + any miscellaneous overhead power.
System Input Current (A) Input current at the operating voltage selected.
System VA Rating (VA) Equal to the System Input Power(W) divided by the power factor (PF).
System BTU/Hr (BTU) Equal to System Input Power(W) multiplied by 3.412.
Server DC Power Consumption (W) DC power consumed by the entire solution operating at the % utilization selected.
Power Utilization % of the power budget consumed when the system is operating at 100% utilization.
System Input Power - Idle (W) Input power consumed when all solution components are in the idle state.  Note, UEFI and operating system must be optimized for minimum idle power.
System Input Power - Off (S5) (W) Input power consumed when the entire solution is in the soft off state.
System Input Power - 100% Stress (W) AC or DC input power.  Equal to the System DC Power - 100% Stress (W) divided by the PSU efficiency + any miscellaneous overhead power.
System DC Power - 100% Stress (W) DC power consumed by the entire solution operating at 100% utilization selected.
Product Carbon Footprint (KG) The PCF value is used to estimate the carbon emissions resulting from the server's electricity consumption over its entire lifecycle, which can be configured in the 'Total Cost of Power Settings' menu on the homepage. Additionally, the 'PCF emission factor' can be adjusted to account for geographical differences across various countries or regions.
Power Policy, Time Duration, Fan Speed, Load Factor Misc options that can be set when the solution is configured.  These options influence the power draw.

20 Power Policies

The most conservative power policy depends on the power policy selected on the server. The total system power may exceed the range of the current power supply.

In general, the total power available to the server is:

  • N+N redundancy with oversubscription: Available power =(1.2) * (PSU rating) * (number of installed PSUs / 2). If a fault occurs on >=N PSUs, throttling could occur. N typically ranges from 1 to 3. For example, if N=3, the server has 6 PSUs and, if 3 or more PSUs fail, there is the possibility of throttling.
  • N+N redundancy without oversubscription: Available power =(PSU rating) * (number of installed PSUs / 2). If a fault occurs on >=N PSUs, no throttling will occur.
  • N+1 redundancy with oversubscription: Available power =(1.2) * (PSU rating) * (number of installed PSUs -1). If a single PSU fails, throttling could occur. N typically ranges from 1 to 5. For example, if N=4, the server has 5 PSUs and, if one PSU fails, throttling could occur.
  • N+1 redundancy without oversubscription: Available power =(PSU rating) * (number of installed PSUs -1). If a single PSU fails, no throttling will occur.
  • Non-redundant: Available power =PSU rating * (number of installed PSUs). If a power supply fails, the server could either be throttled or, under server conditions, shutdown unexpectedly.

Notes

  • PSU =power supply unit
  • Oversubscription has additional power headroom from the redundant PSUs when they are operating under nominal conditions.
  • In cases where throttling can occur, the occurrence of throttling depends on the system configuration and the current workload of the server.