Welcome
TO THE INTEL® XEON® PROCESSOR-BASED SERVER REFRESH SAVINGS ESTIMATOR.
Developed with key learnings from Intel IT's experience in justifying server refresh internally, this tool allows you to enter data about your existing environment and evaluate the benefits of replacing it with the latest generation of Intel-based servers, Intel® Ethernet Products and Intel® Solid-State drives.
TO THE INTEL® XEON® PROCESSOR-BASED SERVER REFRESH SAVINGS ESTIMATOR.
Developed with key learnings from Intel IT's experience in justifying server refresh internally, this tool allows you to enter data about your existing environment and evaluate the benefits of replacing it with the latest generation of Intel-based servers, Intel® Ethernet Products and Intel® Solid-State drives.

Based on Intel IT’s methodology for determining the appropriate server sizing for our scale-up enterprise resource planning (ERP) environment, this tool allows you to enter data about your existing data center environment and evaluate the optimal servers for your project life cycle.

Based on Intel IT’s methodology for determining the appropriate server sizing for our scale-up enterprise resource planning (ERP) environment, this tool allows you to enter data about your existing data center environment and evaluate the optimal servers for your project life cycle.

User Aids:

  • Click or tap the text with a dotted underline to reveal a tooltip.
  • Click the "lock" to reset individual fields to default values
  • Click “Hide All” to hide the inputs under each category
  • Click “Show All” to display inputs for all categories
  • Use the ”See MORE Configuration Options” button to expand your input options

Estimator last updated -.
Exchange rates last updated on -.

Look to optimize your saving via required hardware IT spends - number of servers, SSD, and 10 Gb adapters and possible reallocation of labor resources.

Terms & Conditions

Get Started >

This tool models the combined effects of compute, memory and I/O to understand the overall implications for server sizing and selection to meet IT service level agreements.

It incorporates the required number of servers to address projected workloads, use of SSD vs HDD, migration from 1 Gb to 10 Gb adapters, and the possible freeing of labor resources to address other critical IT business needs.

Reduced operating system licensing costs is one of the reasons for high ROI when consolidating servers. It is important to update the per-server OS licensing cost to reflect your actual situation.


Tool capabilities include:

  • Run a simple analysis with just a few inputs
  • Customize your analysis to be specific to your environment
  • Model up to 5 different existing / new servers profiles
  • Select a variety of on-screen output options to aid in modeling
  • Select what cost & savings variables you'd like to include or exclude
  • Generate detailed reports to share in internal decision making

User Aids:

  • Click or tap the text with a dotted underline to reveal a tooltip.
  • Click the "lock" to reset individual fields to default values
  • Click “Hide All” to hide the inputs under each category
  • Click “Show All” to display inputs for all categories
  • Use the ”See MORE Configuration Options” button to expand your input options

Estimator last updated -.
Exchange rates last updated on -.

Look to optimize your saving via required hardware IT spends - number of servers, SSD, and 10 Gb adapters and possible reallocation of labor resources.

Terms & Conditions

Get Started >

The server refresh scenario is used to determine the proposed method of consolidation. If your current environment is virtualized, select Virtual to Virtual scenario. If your current environment is not virtualized today, but you intend to deploy virtualization on your new servers, select Physical to Virtual scenario. If your current environment is not virtualized today and you plan to remain not virtualized, select the Physical to Physical scenario.

Review and edit information about the existing servers you desire to replace and the new ones you'd like to deploy.

Enter the type of server you would like to refresh with.

Server price, performance per server, power (busy and idle) and software defaults costs will update automatically based on the server type changes as long as you have not overridden those fields. If you have changed one of these listed fields manually, the automatic update feature will no longer work.

To restore the automatic update feature, you must reset all affected fields to the default calculated value by clicking on the "lock" button provided in that field.

Source: Intel® Corporation for Intel-based systems and Alinean, INC for all others.

Review/modify the type of existing servers. Change the model type to evaluate different server refresh situations. Server price, performance per server, power (busy and idle) and software defaults costs will update automatically based on the server type changes as long as you have not overridden those fields. If you have changed one of these above fields manually, the automatic update feature will no longer work.

For the automatic update feature to work, you must reset affected fields to the default calculated value by clicking on the "calculator" button provided in that field.

Estimate the age of your existing servers. The age of the existing servers is used to estimate the total performance of your existing installed base, which is then used to calculate the number of new servers you need to match that estimated performance. The age of your servers is also used in computing depreciation and possible write-off costs.

Server performance based on best publicly available SPECint_rate_base2006* results from www.spec.org as of March 19, 2009. For the following processors, performance results are based on Intel internal estimates: 1 Socket: Intel® Xeon® 3.8GHz, Opteron* 1222, Opteron 1220, Opteron 180, Opteron 150 2 Socket: Intel® Xeon® 2.8GHz, Opteron 250 4 Socket: Intel® Xeon® 7041, Intel® Xeon® MP 3.0GHz, Opteron 2.4GHz 2 core, Opteron 2.4GHz 1 core.

If you wish to use a different benchmark or your own real world application test, adjust the performance results of both the existing and new servers.

Review and edit the server related information for this Profile.

Review/modify the number of existing installed servers and the number of replacement servers needed to support that current environment.

Specify the number of additional servers that are needed at this time to support incremental (new) processing demand.

Any incremental servers are assumed to be of the same server type as the proposed new Intel servers selected above.

Default calculations for the new incremental server demand is calculated below:

Physical to Physical: The new incremental server demand will be assumed to be the same as existing. It is assumed you would add current technology for any incremental computing needs regardless of whether you refresh or not.

Physical to Virtual: The new incremental server demand is reduced by the projected VM loading of the new servers. Example: if new VM loading is 10VM/server, and user adds 50 existing incremental demand, then new incremental demand will be 5 servers.

Virtual to Virtual: The new incremental server demand will be assumed to be the same as existing. It is assumed you would add current technology for any incremental computing and projected VM loading for the new environments.

Adds the number of servers from original servers for replacement and servers added for incremental demand.

Review and edit the amount of average virtualization per server desired for the servers in this Profile.

Review/modify the number of VM/s per server for your existing and new environment. If you desire to update this field but are unable to, ensure that you update your Refresh Scenario to be either Physical to Virtual or Virtual to Virtual. Restore default calculations with the lock buttons

The new VM/server default calculation is calculated as follows.

User may change these values to model their desired virtualization loading per server in the new or old environment.

Virtualization loading will affect the % server utilization calculations below.

Total VMs calculates the number of virtual machines based on the number of servers and VM/server data you have entered.

Review and Compare your modeled total utilized performance before/after refresh.

Utilized Performance is calculated based on the amount of total performance available reduced by the average anticipated utilization of those servers.

Percentage of time server is busy running jobs. Review and compare your modeled total utilized performance before/after refresh. Utilized Performance is calculated based on the amount of total performance available reduced by the average anticipated utilization of those servers.

Default Estimated Existing Server Utilization is assumed to be 10% per Virtual Machine, up to a maximum of 100%.

Default Estimated New Server Utilization Calculation is based on many factors:

The default formula used to calculate the Physical to Virtual and Virtual to Virtual utilization for new servers reads as such:

Replacement Utilization = {(Existing Server Utilization/Existing VM loading per server)*1.15*Replacement VM loading per server}*{Existing Performance per Server/Replacement Performance per Server}

Estimated utilized performance is calculated by taking number of servers times the performance per server (SPECint_rate_base2006 benchmark is default) times the percent utilization for the servers.

This section allows you to edit the software costs for this profile.

Software Licensing Costs represent the cost per server to purchase new licenses for your new environment or any incremental demand for existing environment. For proper costing, ensure this field is cost per server.

Use the Software Wizard for aid in modeling your software costs.

Software maintenance fees represent the cost per server to maintain existing and new licenses per year for your existing and new environments, including any incremental demand for existing environment. For proper costing, ensure this field is cost per server per year.

Use the Software Wizard for aid in modeling your software costs.

The cost per server for validating software configurations on newly installed servers (applies for the new server environment plus incremental demand, if any).

This section allows you to include or exclude a variety of common financial variables in your estimates of Total Cost of Ownership (TCO) and Return on Investment (ROI).

In consolidation or constrained infrastructure scenarios, refresh allows increased compute output (performance, hosts, etc.) with less physical resources. In these cases, the cost of avoiding incremental data center capacity should be when evaluating the benefits of server refresh.

If existing servers are replaced before they are fully depreciated, then this field will allow you write off that residual value as a savings.

Often server refresh enables a smaller physical server environment by running fewer, higher-performance, more energy-efficient servers. This field enables you to include the benefits from reduced power / cooling costs.

Annual software licensing or maintenance fees based on the number of servers, cores, processors that are directly related to the number of servers operated on an annualized basis. Operating a smaller fleet of servers may provide a financial benefit for software fees.

Maintaining aging servers can have costs in extended warranties resulting in higher per server maintenance costs.

Network costs are often related to the number of servers in your environment. Refresh may affect the size of the server fleet resulting in a cost reduction.

For Profit businesses often can write off a portion of their operating costs as a tax benefit reducing total costs.

If you are a NON PROFIT Business, do NOT pay taxes, or your corporate protocols do not include tax implications, we recommend you exclude this variable.

This will add any costs associated with installing new equipment for servers identified for both replacement or incremental compute demand.

This will add any costs associated with validating a new software environment on any new servers you are deploying (either replacement or incremental compute demand).

This will add any costs with removing targeted end-of-life servers in the specified refresh scenario.

This section allows you to specify various initial costs.

System pricing is based on list price for equivalently configured servers from Dell, HP, IBM, and Sun. For current generation Intel-based 1S servers, pricing is based on average price of Dell* PE840, HP* ML110 G5, IBM* x3200 M2 as of March 4, 2009. All systems configured with 1 Intel® Xeon® X3360 or higher processor, 4x2GB memory, 1x160GB SATA drive, 1 PSU. For current generation AMD-based 1S servers, pricing is based on HP* ML115 as of March 13, 2009. For current generation Intel-based 2S servers, pricing is based on average price of Dell 2950 IIII, HP DL380 G5, IBM X3650, and Sun X4150 as of March 4, 2009. All systems configured with 2 Intel® Xeon® X5460 processors or higher, 8x2GB memory, 1x72GB SAS drive, 1 PSU. For current generation AMD-based 2S servers, pricing is based on HP DL385 G5p as of March 13, 2009. For current generation Intel-based 4S servers, pricing is based on average price of Dell R900, HP DL580, IBM X3850, and Sun X4450 as of March 4, 2009. All systems configured with 4 Intel® Xeon® X7350 processors or higher, 16x2GB memory, 1x72GB SAS drive, non-redundant power supply. For current generation AMD-based 4S servers, pricing is based on HP DL585 G5 as of March 13, 2009.

Source: Intel® Corporation for Intel-based systems and Alinean, INC for all others.

The costs to install a new server (cost per server).

The costs to dispose of existing servers (cost per server).

The cost to upgrade network infrastructure as part of the profile. Example: LAN optimization or refresh (chassis/cable, etc) as part of refresh.

The cost to upgrade storage infrastructure as part of the profile. Example: Include additional SAN or upgrade to SAN to complement virtualization and/or refresh.

The cost to upgrade rack infrastructure as part of the profile. Example: upgrading from 1u rack to a blade environment.

Placeholder for any additional costs for the profile evaluation. Example: consulting fees etc.

This section allows you to modify various maintenance costs and warranty information.

Enter the number of years for your server warranty.

Cost per server per year for hardware maintenance/support of any "out-of-warranty" maintenance.

Cost per server per year for network maintenance expense. Since the number of network ports changes with the number of servers, this variable cost is captured as a savings when moving from existing to new server environment.

This section allows you to modify various costs associated with powering and cooling your servers.

Server power consumption is based on internal measurements or estimates from Feb 2008 through March 2009. For current generation Intel-based 2S servers, idle and peak power consumption is based on the average power consumption of Dell* PE2950 III and HP* DL380 G5 configured with 2 Intel® Xeon® X5460 processors or higher, 8x2GB memory, 1x72GB SAS drive, 1 PSU. For current generation AMD-based 2S servers, idle and peak power consumption is based on HP* DL385 G5 and DL385 G5p configured with 2 Opteron* 2352 processors or 2 Opteron* 2384 processors, 8x2GB memory, 1x72GB SAS drive, 1 PSU. For current generation Intel-based 4S servers, idle and peak power consumption is based on the average power consumption of Dell* R900, HP* DL580, IBM* X3850, and Sun* X4450 configured with 4 Intel® Xeon® X7350 processors or higher, 16x2GB memory, 1x72GB SAS drive, non-redundant power supply. For current generation AMD-based 2S servers, idle and peak power consumption is based on HP* DL585 configured with 4 Opteron* 8352 processors or higher, 16x2GB memory, 1x72GB SAS drive, non-redundant power supply configuration.

Server power consumption is based on internal measurements or estimates from Feb 2008 through March 2009. For current generation Intel-based 2S servers, idle and peak power consumption is based on the average power consumption of Dell* PE2950 III and HP* DL380 G5 configured with 2 Intel® Xeon® X5460 processors or higher, 8x2GB memory, 1x72GB SAS drive, 1 PSU. For current generation AMD-based 2S servers, idle and peak power consumption is based on HP* DL385 G5 and DL385 G5p configured with 2 Opteron* 2352 processors or 2 Opteron* 2384 processors, 8x2GB memory, 1x72GB SAS drive, 1 PSU. For current generation Intel-based 4S servers, idle and peak power consumption is based on the average power consumption of Dell* R900, HP* DL580, IBM* X3850, and Sun* X4450 configured with 4 Intel® Xeon® X7350 processors or higher, 16x2GB memory, 1x72GB SAS drive, non-redundant power supply. For current generation AMD-based 2S servers, idle and peak power consumption is based on HP* DL585 configured with 4 Opteron* 8352 processors or higher, 16x2GB memory, 1x72GB SAS drive, non-redundant power supply configuration.

Cost per kilowatt hour for utility power.

Factor to estimate future utility rate increases year over year ($/kWh).

Estimate the approximate average hours of server operation for this profile.

Power Use Effectiveness (PUE) is a factor to estimate infrastructure cooling efficiency and cost requirements. The default value is set at 2.0. A PUE of 2.0 means that your non-IT power (e.g. cooling, HVAC, lighting, etc.) is equal to your IT power (e.g. server, storage, networking). Therefore total power consumption is estimated 2 times the IT power consumption.

Estimated Cost (in $ per KW) for Data Center infrastructure capacity additions. This field is used to calculate the cost avoidance savings of not having to build new server infrastructure to accommodate incremental growth if you did not refresh your old servers.

This section allows you to modify various financial parameters.

Weighted Average Cost of Capital or Hurdle Rate for Organization.

Organization's marginal tax rate.

This section allows you to address the cost of depreciating hardware.

This field allows you to adjust the use life for straight line financial depreciation for any new servers purchased in this profile.

This field allows you to adjust the use life for straight line financial depreciation for any network implementation costs identified in this profile.

This field allows you to adjust the use life for straight line financial depreciation for any storage implementation costs identified in this profile.

This field allows you to adjust the use life for straight line financial depreciation for any rack implementation costs identified in this profile.

This section allows you to capture other costs in addition to what have entered already that apply to your profile.

Enter any costs taken today to offset costs in the future, such as replacing parts before they fail, etc.

Enter any other costs you may have not yet entered.

If applicable to your environment.

This section allows you to specify migration details from HDD to SSD.

Form factor is the size of the HDD in your environment.

Storage interface is how the HDD communicates with the host system. It impacts price of the solution.

RPM impacts performance, power consumed, and price of the storage solution.

Total capacity impacts price of the storage solution.

The number of installed HDDs impacts performance, power consumed, and price of the storage solution as well as replacement and maintenance costs.

Enter a percentage that represents the amount of capacity actually used in each HDD in your environment. (To obtain better performance, some systems reserve free space on each HDD. For example, a system requiring 1 TB of storage may deploy 10 TB of HDDs, using only 10% capacity on each HDD.) This option is only required on systems with “No RAID” configuration.

The type of RAID configuration used impacts HDD performance and the actual capacity required for the storage environment.

Choose the type and density of the Intel SSDs. Over-provisioning refers to allocating free area in the SSD to increase random mixed workload performance. It is recommended to start with a non-over-provisioned configuration.

The desired RAID configuration of the target systems.

This section allows you to specify existing labor details to understand potential labor savings.

Number of engineering employees in current environment.

Average salary of engineering employees.

Number of operations employees in current environment.

Average salary of operations employees.

Number of administration employee’s in current environment.

Average salary of administration employees.

Number of other employees (not fitting into the above categories) in current environment.

Average salary of other employees.

The savings per dollar believed possible based on existing environment. Ex: 100% savings factor would mean you would save one dollar for every dollar of possible savings while a 50% savings factor would mean $0.50 savings realized on every $1.00 of possible savings. This number is based on the efficiency and automation that exists in the current environment. A less automated business would realize a higher savings factor than a more lean/efficient business.

This section will assist you in assessing the benefits of moving from 1 Gb adapters to 10 Gb adapters for the servers in your data center.

Number of Ethernet ports (adapters) per server in the existing state.

Number of Ethernet ports — 10 Gb or 1 Gb — (adapters) in the new state.

Chose the desired adapter type.

Do you want to connect to the SAN via iSCSI, FCoE, or Fibre Channel port or not.

Enter the number of Fibre Channels.

Select the speed of your Fibre Channels.

Enter the number of Ethernet ports.

Select the type of adapter card.

The Baseline server is the server configuration you are most familiar with for the application workload you are modeling. The tool compares the Baseline server data against data entered in the Advanced mode to create a comparison for future utilization needs.
CPU Description provides a simplified configuration for the processor being modeled. It includes the number of processors supported (2P = 2 processor, 4P = 4 processor, etc) and the specific processor model being used for the selected server configuration. The description is simplified for display and provides only the frequency, number of cores, and cache size that are unique to that processor model. For more detailed information about Intel Xeon processors visit http://ark.intel.com.
CPU Performance is an estimate of the server performance for the modeled workload. The modeling tool defaults to specINTRate 2006* - a benchmark commonly used as a measure of performance for a variety of compute intensive workloads. If desired, you can replace these defaults with measures more appropriate for your workload and environment. REMINDER: if you change the performance benchmark for one server configuration (i.e. Baseline), you must also change the performance for the other 2 server configurations in the Advanced mode. This is necessary for the tool to properly project estimated utilization numbers for those server configurations.
The memory capacity in GB installed on the selected server.
The number of I/O slots available on the selected server.
Initial Average Utilization represents the average consumption by resource type of your modeled workload on the reference server.
Alternate A is the first alternative server configuration to be compared to the Baseline server for your projected workload.
The processor configuration for the selected server - this will specify the number of processors as well as the processor model. The processor model uniquely identifies the selected processor type which maps to specific number of cores, frequency, memory bandwidth and cache size that are unique to that processor model.
CPU Performance is an estimate of the server performance for the modeled workload. The modeling tool defaults to specINTRate 2006* - a benchmark commonly used as a measure of performance for a variety of compute intensive workloads. If desired, you can replace these defaults with measures more appropriate for your workload and environment. REMINDER: if you change the performance benchmark for one server configuration (i.e. Baseline), you must also change the performance for the other 2 server configurations in the Advanced mode. This is necessary for the tool to properly project estimated utilization numbers for those server configurations.
The memory capacity in GB installed on the selected server.
The number of I/O slots available on the selected server.
The initial average utilization for alternate servers A and B is calculated based on the initial average utilization for the reference server and the relative capacities of the alternate servers (relative to the reference server).

For instance, if alternate server A has double the compute capacity of the reference server, then the initial average CPU utilization of alternate server A will automatically be set at half the value on the reference server.

Conversely, if the initial average CPU utilization on the reference server is increased to be double the original default average utilization setting, then the initial average CPU utilization of the alternate servers will automatically be doubled as well to reflect the increased intensity of the workload.
Alternate B is the second alternative server configuration to be compared to the Baseline server for your projected workload.
The processor configuration for the selected server - this will specify the number of processors as well as the processor model. The processor model uniquely identifies the selected processor type which maps to specific number of cores, frequency, memory bandwidth and cache size that are unique to that processor model.
CPU Performance is an estimate of the server performance for the modeled workload. The modeling tool defaults to specINTRate 2006* - a benchmark commonly used as a measure of performance for a variety of compute intensive workloads. If desired, you can replace these defaults with measures more appropriate for your workload and environment. REMINDER: if you change the performance benchmark for one server configuration (i.e. Baseline), you must also change the performance for the other 2 server configurations in the Advanced mode. This is necessary for the tool to properly project estimated utilization numbers for those server configurations.
The memory capacity in GB installed on the selected server.
The number of I/O slots available on the selected server.
The initial average utilization for alternate servers A and B is calculated based on the initial average utilization for the reference server and the relative capacities of the alternate servers (relative to the reference server).

For instance, if alternate server A has double the compute capacity of the reference server, then the initial average CPU utilization of alternate server A will automatically be set at half the value on the reference server.

Conversely, if the initial average CPU utilization on the reference server is increased to be double the original default average utilization setting, then the initial average CPU utilization of the alternate servers will automatically be doubled as well to reflect the increased intensity of the workload.
Select from a variety of output views to aid in decision making and modeling. By default, projections for all 3 server configurations are shown in chart form. To compare only 2 servers, uncheck the configuration you wish to ignore. To see more detail on resource utilization projections, click the appropriate magnifying glass. To see detailed utilization projection data, switch to table view.
Check the server profiles you would like to include in the charts and tables.