Digitization is driving the need for increased resiliency across data center architectures, from cloud-based data centers to the edge. I co-authored a white paper a couple of years back focused on this very topic. But what does it mean to be resilient? How do I ensure my business processes remain available? In the past, most would turn to the various existing tier or criticality methods to assess a data center or design a new data center to the required availability level.
About 2 ½ years ago, The Green Grid saw an opportunity to build on these methods, not only to contemplate traditional redundancy and availability approaches, but to also consider the impact to availability by innovative and sustainable energy. I have had the privilege of participating in the workgroup that collaborated on an open standard for data center availability, called OSDA. The output of this workgroup resulted in an online tool assessing availability levels. It is a work-in-progress as it focuses mainly on the electrical infrastructure right now, but even in its current state, can offer some interesting guidance.
Limitations of Availability Ratings Prompted a New Approach
Most data center designers and operators are familiar with the common availability or tier ratings that exist on a scale of 1 to 4. In fact, my colleague Victor Avelar wrote a white paper several years back on Specifying Data Center Criticality / Tier Levels, where he analyzed different existing tier methods and proposed a criticality spec. The organizations that pioneered these rating systems were instrumental in providing the data center industry with a reference for how to classify data center designs (basic non-redundant, basic redundant, concurrent maintainable, and fault tolerant). These are still the go-to standard in evaluating detailed designs and obtaining 3rd party certifications. But do these methods fall short as we look at new ways to include sustainable energy in data center design? These approaches:
- are on a scale of 1 to 4 which doesn’t offer enough differentiation of designs
- are sometimes overly prescriptive in system design requirements
- don’t consider the availability of the utility source
- don’t include options for renewable energy
- focus on a single site’s availability, not the business application(s)
- are often proprietary, making it difficult or costly to know if you’ve met the criteria for a particular level
These limitations became the focus for the OSDA workgroup.
Background on the Development of OSDA
OSDA aims to be a “less prescriptive framework”, one that contemplates the impact of alternative energy sources and innovative designs focused around efficiency, to be used in the early planning phase of a data center project. As the name emphases, it is “open”, meaning it is a non-proprietary standard, and the group continues to look for input and cross-industry collaboration so that designers can weigh the tradeoffs of all design choices, and have a more holistic view of the data center.
The tool was developed on a 10-point relative scale and is based on 3 types of calculations: (1) probability distribution function to model the frequency and duration of outages of energy sources; (2) Monte Carlo simulations to model the availability of the combined components into subsystems; and (3) reliability block diagrams to model the interconnection of the energy source and the subsystems, as well as offsite redundancy.
Try the OSDA Tool Yourself
While there’s some sophisticated data and math on the back-end, the front-end interface is intended to be simple, so you can rapidly understand the availability tradeoffs in data center design.
Source: The Green Grid
You can see that some of the modules are “in development” (the greyed-out blocks). As the tool gains traction and interest, the vision is for future releases to expand in scope and consider all aspects of a data center, such as: water source, telecom and networking, as well as more sophisticated cooling modeling. There are resources referenced in the tool if you want to learn more about the methodology and assumptions.
Assess the Availability of Alternate Energy Sources
Let’s say I’m deploying new IT equipment to hundreds of edge sites. Resiliency is critical, but I don’t have a generator available at many of the sites. What’s the impact on my availability? What alternatives can I implement to achieve the same level of availability as a generator would have provided me? OSDA can help answer this.
In the first screenshot below, I selected an N+1 standby generator for my typical grid utility energy source. My subsystem energy source OSDA score was a 6.3. But what’s my OSDA score when regulations prevent me from using a generator? The second screenshot shows the same utility source but no standby generator. With 3 hours of energy storage (i.e. li-ion batteries), I can achieve the same level of availability.
Source: The Green Grid
You can also use the tool to see how wind or solar energy sources impact the results. This is increasingly important as energy storage and alternative energy sources continue to grow.
If you haven’t checked out the OSDA tool yet, I encourage you to do so. You’ll see that there are alternative energy efficient approaches to achieving the availability levels you’re targeting.
Leave a comment below or check out other blog posts from the Data Center Science Center team.