![Industry Perspectives Industry Perspectives](https://eu-images.contentstack.com/v3/assets/blt8eb3cdfc1fce5194/blt79a1824cfad941c4/664ba129be90a086233eb8e0/Industry_Perspectives_DCK_light.png?width=700&auto=webp&quality=80&disable=upscale)
Insight and analysis on the data center space from industry thought leaders.
Forecasting to Improve Your Data Center Portfolio
Building and managing data center infrastructure represent large amounts of investments, writes Santiago Bernal of Microsoft. Knowing if and when you need additional capacity can translate into a significant improvement to your data center portfolio.
December 10, 2014
![Forecasting to Improve Your Data Center Portfolio Forecasting to Improve Your Data Center Portfolio](https://eu-images.contentstack.com/v3/assets/blt8eb3cdfc1fce5194/bltdd8212fb18af8c17/662123116cb9f075e620767e/Santiago_Bernal_TH.png?width=1280&auto=webp&quality=95&format=jpg&disable=upscale)
Santiago Bernal (CSCP) is a Senior Project Manager in the Cloud and Enterprise team at Microsoft Corporation.
Building and managing data center infrastructure represent large amounts of investments; it can easily reach up to hundreds of millions of dollars. Knowing whether or not you need additional capacity or whether your company can wait a few months to make this additional investment, can translate into a significant financial improvement to your data center portfolio.
In retail, there are two forecasting models that allow you to manage your inventory levels between Original Equipment Manufacturers (OEM), distribution centers, retail stores and end customer: sell in and sell through models.
Sell In Forecasting Model: allows you to determine your inventory needs based on how much inventory your company has at your retail stores. Based on re-order points, additional inventory is requested from distribution centers or the OEM.
Sell Trough Forecasting Model: ensures that inventory in the channel is replenished when your existing inventory is sold to end customers. This type of model can be seen as a deeper dive into your current existing inventory. Is your inventory being sold to your customers? Or is it just moving from the stock room to your shelves on the floor?
By applying these retail principles, we can ask ourselves the same question regarding the number of data centers and servers that a company has deployed to meet customer demand. Is your current server fleet meeting actual customer demand (selling though) or are they just being installed to meet future (sell in) forecast? Is your company solely looking at how full (servers installed or sell in) your data centers are or is it looking at the utilization of these servers (sell through)?
Sell In = Servers installed at your data center
Sell Through = Server utilization at your data center
Let’s simulate these principles with the scenario below:
Company ABC has three data centers in a specific global region. Based on the lead-times built into their re-ordering point model, this company will trigger the build of another data center when sell in reaches 80 percent at a regional level. The percentages of sell in for each data center are illustrated below:
Based on additional sell in demand from customers, this company will trigger the signal to build a fourth data center in this region because this additional sell in demand will reach 80 percent regional sell in target. This additional investment will translate into a $100 million investment.
Company XYZ has a similar data center portfolio and has decided to include sell through into their forecast and inventory management logic. Based on the lead-times built into their re-ordering point model, this company will trigger the build of another data center when sell through reaches 70 percent at a regional level.
Based on additional sell through (actual server utilization), this company decides not to trigger a build for a fourth data center in this region (even though the sell in target has been reached). This decision saves Company XYZ millions of dollars.
Company XYZ has also given themselves a few more months to analyze the age of their current inventory on hand before having to make a decision to acquire additional capacity. During this period, Company XYZ determined that the servers in the first data center are approaching the end of their life cycle and will therefore be decommissioned. These servers will need to be retired and replaced with new, more efficient servers and network designs. Company XYZ determines that data center one can now be used to home the new and more efficient network and servers instead of having to invest in a fourth data center.
This analysis and scenario is described below:
Company ABC invested in the fourth data center, not realizing that the inventory in data center one is becoming obsolete. Sell in for data center four has now increased, but sell in for data center one is now at 0 percent as the servers residing in this data center have now reached their end of life.
Company XYZ decided not to invest in data center four as they now have capacity available in data center one.
To be able to implement this type of additional forecast model, you will need to invest in automation so actual utilization is able to be measured accurately and recorded in real time. Inability to do so could add risk to your current data center portfolio growth plans. In addition, the implementation of a basic Sales and Operations Planning (S&OP) process will allow you to have direct conversations between your customers, finance, engineering groups and operations. Yes, “Forecasts are ALWAYS wrong,” but additional investments in getting additional details about the actual utilization of your servers will translate into a big pay day.
Adding focus to sell through (server utilization) forecasting in addition to sell in, can improve your data center portfolio performance.
Industry Perspectives is a content channel at Data Center Knowledge highlighting thought leadership in the data center arena. See our guidelines and submission process for information on participating. View previously published Industry Perspectives in our Knowledge Library.
About the Author
You May Also Like