Case Study: TSO LogicThe premise was easy enough. We don't leave motor vehicles idling when we aren't using them. Why couldn't we apply the same logic to computers in server farms.
Data centres are built to support their busiest day or peak capacity. But what happens during the rest of the time? The IDC [International Data Corporation] estimates that 75% of servers that are powered on are actually not being used. That was the simple question that had perplexed Aaron Rallo, Director, President and CEO of PNI Digital Media.
User + Audience
One of the initial challenges was defining a persona that the project would serve. Was it the CEO of the enterprise looking to have an ethical approach to energy use; was in the Service Delivery Manager having to ensure “100% uptime”; or perhaps a department head looking to have greater transparency in the cost of computing power for their team.
Once the software was deployed (including the algorithm that was at the heart of the platform), we realized that the product could serve all the above personas. For beta version though, we focused on an a datacentre manager who needed to ensure that performance wasn’t impacted by the decisions the TSO Logic wanted to make.
Roles and Responsibility
My role was the UX Strategist and Designer. As the sole UX+UI designer for the platform, I worked with the agile engineering team to transform its rough WinApp beta version to the commercially available product it become. I was directly involved in brainstorming sessions with Aaron and our product manager as the product matured and we got feedback from our customers.
On the visual design side, I was responsible for creating the suite of marketing materials, trades show materials and the like.
The Process and My Involvement
The first real world installations were with 2 visual effects companies in Vancouver. After running the software for about a month to measure the baseline, we were able to demonstrate over 55% in power savings equating to around $80,000 in annual savings.
With this validation, we knew that we had to turn our attention to the UI and to make it more accessible.
One of the first tasks was to align on taxonomy so that all stake holders had the same uniform understanding of the different states a device could be in. A rubric was created to show how a device transitioned from one state to another.
This naturally lead to the crafting of the design language for the application. Icons were increases in size so that they could convey more information at a glance. The classic traffic light metaphor was used to indicate various states.
One of the challenges for the application, was the load time for the server room hierarchy (and the real time data). I proposed to present a dashboard upon the software being started. The user would instantly get insight into the data centres performance (last 7 days and last 30 days). Information included the number of devices in the network, how many of those were running TSO Logic and the aggregated total of Workload, Capacity, Performance and Power usage.
With continued positive feedback from our 2 test sites, we realized that the next iteration of the software would need to be agent-less and be able to be deployed remotely … the solution had to leverage web technology (as opposed to being platform based).
Much of the original software was refactored and core features were rewritten as a WebApp. This allowed me to explore more intuitive designs, improve how users would interact with the information being presented and also create space for additional features.
Creating a navigation pain removed visual clutter and helped the user access the additional features including reporting, system health and settings. The global kill switch was always accessible from the top bar.
The hierarchy could also now handle complex and adhoc grouping of devices and virtual servers. A summary of nodes was also made available at easy glance.
Additional inputs such as the cost of hardware and software licenses were also introduced to give an even greater sense of the actual cost of computing.
Outcomes and Results
The real breakthrough for TSO Logic came from an unsuspected source. Cloud computing. What enterprises were actually looking for, was a way to right size their computing needs as their business requirements evolved. Rather than replacing obsolete machines on site, a combination of onsite and cloud computing was becoming a viable alternative.
No one cared for saving money in the way we thought. While we could demonstrate making server farms more power and cost efficient through elegantly powering up and down idle devices. Enterprise service level agreements just couldn’t risk having machines offline.
Rather than focus on cost savings, TSO Logic pivoted to “data driven analysis of the total cost of ownership and the cost modelling”.
TSO Logic would provide the base line for current on premis computing requirements (and cost), and then map them to the best fit from the main cloud computing service providers such asGoogle, Amazon Web Services and Microsoft.
This base line was intentionally agnostic its approach. As more of our software was deployed, it began to attract the attention of the very service providers we were recommending. In December 2019, TSO Logic was acquired by Amazon. It’s been renamed to ASW Migration Evaluator, but its core functionality, UX and UI is the same.
My Key Learnings.
Through countless iterations, thousands of hours of development, passionate advocacy by Aaron and his team, we turned a simple premise (to Turn Shit Off) , into one of the worlds premier enterprise tools.
Working within an agile environment really showed me how critical it was to align stakeholders across the project. Learning from our experiments, small incremental improvements all contributed to the critical break though moments. Being tied to legacy code forced me to thrive inside a box and to make the best of what was available.
Once free from those constraints, the core intention remained the same. We could iterate faster and bring a polish to the product to make it ready for mass market.
Taking complex data, transforming that into information that a luddite could use was the quintessential UX challenge. It instilled in me the simple premise to tactfully ask the “why”. Uncovering the core motivation and building upon a a focused intention sets up an environment for success.
I am chuffed to see that the software that I was tasked to reimagine for TSO Logic, still retains its overall functionality and user interface as part of AWS. Check out the promo reel below.