eBay manages billions of listings and millions of transactions daily, so it needed real-time insight into its data center operations to keep systems running and prepare for future demand. By integrating a Data Center Infrastructure Management (DCIM) tool with its AVEVA™ PI System™ data, eBay created a unified “cockpit” view of data center operations, enabling accurate forecasting, improved power and cooling visibility, and faster, data-driven decisions.
Challenges
Lack of holistic, real-time view of data center operations
Reliance on outdated and disconnected spreadsheets
Inability to forecast capacity needs with confidence
Results
Increased insight into power and cooling status and the ability to predict future capacity needs
Faster, data-driven decisions on million-dollar infrastructure investments
Improved forecasting accuracy for equipment and space planning
Every five seconds, a smartphone is purchased on eBay. Every two seconds, a trading card. With 170 million active buyers on its platform and over 1.1 billion product listings on its website, eBay’s data centers are critical to tracking and managing a constant stream of transactions.
Jeff Tepfer, principal architect at eBay, and his team of engineers monitor the core data centers that run eBay’s website, ensuring that they have sufficient server capacity to handle eBay’s daily web traffic. “In reality, [the data centers] are our business,” Tepfer said during his presentation at PI World 2018. “If the data centers do not have the capacity, do not have the availability, our business is offline.”
Solution
A Data Center Infrastructure Management tool that uses AVEVA™ PI System™ data to consolidate infrastructure monitoring, enabling real-time visibility, life cycle planning, and predictive insights.
What’s the forecast?
eBay’s capacity needs within its data centers fluctuate daily, and it has a constant revolving door of equipment coming in and out—500 racks of outgoing and newly incoming gear each year. It is crucial to its business to have a clear understanding of when new equipment comes in and when decommissioned equipment goes out and what corresponding changes to power distribution and cooling are needed.
Additionally, Tepfer and his team need direct insight into the capacities and functionality of its data centers, and quick access to real-time demands, equipment allocation, space reservations and budget forecasting. “If my boss says, ‘Hey, are we going to have capacity in three years?’ I can’t just say ‘I hope so. Maybe. I don’t know.’ It’s a couple million dollars. We have to have tangible answers,” Tepfer explained.
eBay collects and stores 3 million data points in the AVEVA PI System, but much of its data management relied heavily on spreadsheets made by many different people, often with links to other spreadsheets that were either dead or out of date. eBay needed a more scalable and sustainable way to manage and analyze its data, something that would fully leverage the data in its AVEVA PI System.
Clearing the clouds with clarity
A new partnership with Maya Heat Transfer Technologies offered eBay a fresh look at its AVEVA PI System data. A simulation company based in Montreal, Maya develops computer-aided engineering simulation tools and does predictive analysis through machine learning and AI. Maya relies on the AVEVA PI System as its real-time monitoring platform for all its IoT software development projects. Maya’s platform Clarity was the Data Center Infrastructure Management (DCIM) tool that eBay was looking for, combining infrastructure management with real-time monitoring of IT and facility assets that operate in a data center.
First, eBay consolidated all its AVEVA PI Servers and started from the ground up on a new asset framework structure. Once organized, it was able to build a “cockpit” view of its AVEVA PI System data using Clarity. The team created dashboards to monitor things like power allocation in its data centers over time. With its AVEVA PI System data informing its new DCIM, eBay could see which busways were loaded or not loaded and which rows of servers might have additional power demands or be overloaded. The team could also monitor environmental data like temperature to see if there were any cooling issues.
Combining real-time AVEVA PI System data with Clarity’s lifecycle planning ability also allowed eBay to increase the accuracy of its equipment use forecasting. eBay can now look into the future to see when the capacity of racks will need to be reduced or increased. With his new increased insight and forecasting ability, Tepfer can now give his boss direct answers to pressing million-dollar questions about future capacity.
“We needed to make very critical, timely, and expensive decisions in real time.”
—Jeff Tepfer, Principal Architect, eBay
Datacenter Clarity LC ®, powered by AVEVA PI System, enhances business visibility into the real-time power, cooling and environmental conditions of eBay’s data centers.
AI on the horizon
Going forward, eBay and Maya will be looking to apply AI techniques to its AVEVA PI System data. To do this, Maya will use AVEVA PI System’s Event Frames to pinpoint specific events that have occurred during the day-to-day operation of an asset. Information about key events will be used to train algorithms. “Obviously applying AI techniques to data centers has benefits,” Patrick Lepage, senior application engineer at Maya Heat Transfer Technologies said. “The most obvious is the ability to detect a variety of anomalies like equipment failure, electrical spikes and so on. In addition, the fact that we are able to align that PI real-time data with the information we collect from Clarity really provides a much more holistic perspective.”
Product highlights
AVEVA™ PI System™
Collect, aggregate, and enrich real-time operations data for immediate problem-solving and easily deliver formatted data to enterprise applications and advanced analytics.
AVEVA PI Server
Learn about the trusted, high-volume, and real-time data storage, contextualization, analytics, and notification engine at the heart of AVEVA™ PI System™.
Related success stories
Stay in the know: Keep up to date on the latest happenings around the industry.