Advantages of using reference designs

It is no longer practical or cost-effective to completely engineer all aspects of a unique data centre; re-use of proven, documented subsystems or complete designs is a best practice for both new and upgrade data centres. Adopting a well conceived reference design can have a positive impact on both the project itself, as well as on the operation of the data center over its lifetime.

Strong incentives exist to plan, build, and commission data centres faster. However, the reality of tighter budgets, fewer staff, uncertain and dynamic IT loads, and an uncompromised need for high availability makes it difficult to do this optimally. Effective planning is the key because mistakes can magnify and propagate through deployment phases, resulting in delays, cost overruns, wasted time, and ultimately a compromised system. A data centre reference design is a tool used during this planning process that helps avoid some of these potential pitfalls.

Reference designs are built upon recommended and proven best practices. Electrical component manufacturers often provide reference designs for their products to help their customers be more efficient and effective in their application. House builders use models and existing architectural drawings as reference designs to not just highlight their overall capabilities, but also to encourage use of a proven and standardized construction architecture that can save time, effort, and money for both the builder and the homeowner. A data centre reference design acts as a starting point offering a project team several valuable benefits including facilitating and simplifying the planning phase; reducing time to create buildable designs; reducing risk, offering predictable performance and improved reliability of the data centre once operational.

Elements of data centre reference design

A data centre reference design is a tested, validated, and documented plan for how the physical infrastructure systems are to be constructed and laid out, as well as for describing which specific components are used. For a reference design covering a complete data centre, the electrical, mechanical, and IT room areas should all be covered by the plan. The design typically consists of graphical descriptions supported by written documentation.

Perhaps the most useful of all the possible design documents are the engineering “one-line” (or “single-line”) diagrams for the electrical, mechanical, and IT rooms. These documents provide detailed information about the architecture, the number and type of components installed, the flow of power and water through the various systems and components, as well as showing the electrical and plumbing connections to be made. It’s these drawings that largely enable a reference design to be eventually buildable.

Floor layout drawings are another important graphical design description document.

As their name implies, they show the physical dimensions, arrangement, and clearances of all the system components. These components include such things as racks, PDUs, CRACs/CRAHs, UPSs, generators, switchboards, ATSs, etc. This critical information will quickly help answer big questions such as, “can I fit a 1 MW data centre in my building?” Beyond conveying overall space requirements for all the various aspects of the design, the component and system layouts also are oriented in specific ways for specific reasons including maximizing cooling efficiency, simplifying operations, maintaining good security and properly applying codes and standards related to spacing.

3D spatial views are found in some reference designs and provide a clear, high-level overview of what a given reference design will look like once built. All of the major system components can be easily identified and described. Although not necessarily useable from an engineering standpoint, these 3D views can be an effective way to help demonstrate and promote a design to non-technical project stakeholders such as a CFO or supply chain manager.

An effective reference design also includes a high level summary of the design’s attributes and system-level performance specifications. This should list basic specifications such as estimated cost, IT load capacity, estimated annualized PUE, density capability, level of redundancy for power and cooling systems, heat rejection method, and so on. This standardized list of system attributes and performance specs helps project teams to make fast and easy comparisons of designs. Also helpful when comparing designs is to have a document which explains the philosophy and specific benefits of a given design. This information can help explain why elements of the design are what they are; details that might be overlooked or misunderstood otherwise.

A bill of materials (BOM) is another important and helpful element of a data centre reference design. The BOM is a detailed list of all components that comprise what is shown in the reference design graphical documentation. The specific product name, description, part number, and quantity for each part are all identified. And in the example shown below, it is noted whether the reference design vendor supplies the part or not. Having this information early in the planning phase can help save time and reduce risk in the specification and procurement steps of building a data centre.

Any limitations?

Reference designs can convey a lot of detailed design information and provide a significant head start in the overall project. However, to have a full and accurate picture of what a reference design can offer, it is important to understand their limitations and boundaries. Firstly, before a project team can even begin to compare and contrast reference designs, they need to develop and agree on a list of basic requirements for the new data centre, at a minimum based on three inputs: Total expected IT load; Level of redundancy for power and cooling; Density.

With these three values the appropriate reference designs can be selected from a larger library of options. From there, designs can be compared and trade-offs analyzed. Defining a larger, more comprehensive list of requirements, however, can further simplify and more quickly narrow down the number of possible design options.

The benefits of using a reference design

The high level benefits of using a data centre reference design as described above include facilitating and simplifying the planning phase; reducing time to create buildable design; and reducing risk, offering predictable performance and improving reliability of the data centre once operational.

Conclusions

Using a reference design to plan a new data centre saves time and effort while reducing risk and improving reliability. Starting a project with a clear and standardized menu of design options facilitates the planning process. It does so by using a common language to help align goals, encourage cooperation and participation across multiple functions, and makes it easier to evaluate the trade-offs between design goals.

The contents of a reference design such as one-line engineering drawings, a BOM, and detailed dimensions provide a head start in getting to buildable, legal, and localized design plans.

This head start shortens the design cycle and likely saves money in the process. Being built on pre-engineered systems intended to interoperate with each other, reference designs reduce risk and improve overall predictability and reliability as compared to using a custom, one-off design built from scratch.

Using a reference design as a starting point in the planning process helps bring a new data center online faster and helps ensure there are fewer surprises and problems once it’s operational.