3 Real-World Use Cases of Driving Data Center Automation via Integration
Data center managers are increasingly looking for ways to automate tasks to save time and improve data accuracy.
In our recent Automation Workshop webinar, data center experts from eBay, MacStadium, and the University of Chicago shared their real-world use cases and insights on how they use APIs and integration to drive data center automation.
Here are the key takeaways from their presentations as they demonstrated how Data Center Infrastructure Management (DCIM) software dramatically simplifies data center management.
1. eBay Automates Power Budget Calculations and Updates
Before they deployed a modern DCIM solution, eBay had difficulty identifying their power utilization for capacity planning purposes, especially for new devices and new applications. They would seek input from manufacturers, hardware engineering teams, peers in their core data centers, and application owners, but they could not get entirely reliable information. In the case of colo providers, they could only get cabinet-level power utilization upon request and only for a single snapshot in time.
Then, when eBay was building out their DCIM into all their locations, they leveraged their tool’s Auto Power Budget feature. Auto Power Budget is a machine learning algorithm that automatically calculates power budget profiles for each device instance based on how they are used in their actual environment.
“Auto Power Budget is a pretty simple thing to get started on,” said Ken Torres, Global Data Center Engineer. “It’s simply just turning it on.”
eBay configured Auto Power Budget policies that included the day and time they wanted the power budget profiles to be updated and a 20% safety margin of the maximum power utilization.
Auto Power Budget polls each outlet on eBay’s intelligent rack PDUs every 10 minutes, so they have highly accurate data as to what the load of each outlet is for every power supply in their environment.
“Over the course of the week, that’s about 10,000 times [per outlet],” Torres said. “It’s a very granular set of data.”
Torres demonstrated one device as an example of the value of Auto Power Budget.
“The original power budget of this was 450 watts, and Auto Power Budget has adjusted it down to 45% of the nameplate which means it’s 339 watts. While that doesn’t seem like a big deal, it is to us because some of these deployments are 100 servers. If you take the difference between the original power and what’s been budgeted, that’s about 10 kW. That’s not a small amount of power.”
Using these new power profiles, eBay uses their DCIM software’s intelligent capacity search feature to find the ideal cabinet to place their servers in and what-if analysis to see their projects’ impact on capacity.
Ultimately, Auto Power Budget enables eBay to get more compute power with less capacity and less floor space.
Torres gave a specific example of how this saves eBay money:
Prior to having Auto Power Budget, eBay had a project that their calculations determined would require six cabinets in each of six locations. This year, with a similar project, they have learned via Auto Power Budget that it will actually only require four cabinets in those six locations.
“A cabinet–between the cabinet, power supplies, DCIM integration, power installation, and installation itself—is about $10,000 to us. So just for this very project… that’s a cost avoidance of about $120,000. It’s pretty significant. It’s one of those nice things you like to raise up to your directors or senior managers.”
eBay is also driving automation by integrating their DCIM software with ServiceNow.
“A number of times for projects, we go build all this into our DCIM, but we don’t know what the serial numbers or the asset tags are going to be at the time we build this,” said Torres. “Rather than us revisiting this, we’re going to use the ServiceNow integration so that when our operations team actually does add the serial numbers and asset tags, it’ll automatically populate. That saves a pretty big step for both our teams.”
Mark Jones, Global Data Center Engineer, added, “Say we didn’t put a rack in our DCIM. If something is installed from the ops team, ServiceNow will actually go and tell our DCIM which cabinet and which U, and will automatically place the item in our DCIM when it’s been inputted into ServiceNow.”
2. MacStadium Automates Back-Office Processing
MacStadium is always working to improve the online provisioning experience for both internal and external customers. They want to provide fast fulfillment to external customers while meeting or exceeding SLAs, and they accomplish this by having high-performance back-office operations.
MacStadium needed a solution to help manage the high volume of changes they experience without disrupting existing customer experiences or workflows of internal teams. They also wanted to eliminate manual data entry and the possibility of human error. A modern DCIM solution with open APIs was just what they needed.
According to Robert Perkins, Lead Infrastructure Engineer/Architect, “We can assign slots and compute assets to be ready on demand for customers signing up on our website. Our automation automatically moves compute assets from our pre-staged account directly to our customer’s accounts and updates all our internal systems to reflect the completed asset under the customer’s account.”
MacStadium also uses APIs to streamline operations with customers who order colocation services.
“Using the API, we’re able to pull our existing racks, see where we have space available, and assign that space automatically to a customer order via our website so they can send in their equipment,” Perkins said. “We sync the rack slots, switch port, PDU port, make and model, serial number, and asset ID between all of our systems. This reflects changes across our billing platform, customer portal, administration system, and accounting systems all pulling data from our DCIM software.”
“Using the API,” Perkins continued, “our team can utilize the same workflow we had in place before, but we’re getting all the benefits of DCIM and being able to view our data centers remotely. We have global data centers, and we can’t always be there to see what changes need to be placed. And that’s what’s great about DCIM. The 3D visualization allows us to be there and is our full source of truth for all our data centers.”
“It’s been a great tool for us to allow our executive team to see our facilities virtually,” said Perkins.
3. The University of Chicago is Making DCIM the “Center of the Universe”
The University of Chicago—like many organizations—has many different tools and systems, and they need to have them all communicate to fully understand what is going on in their environment.
“Without data, you can’t track or figure out what you’re doing,” said Raymond Parpart, Director Data Center Strategy & Operations.
The University of Chicago has been working on deploying a variety of monitoring tools, and they are making their DCIM software the “center of the universe for the data center.” As such, it needs to integrate with many different systems.
For example, the University of Chicago has integrated their DCIM software with Slack and Teams.
“We’ve created a Command Center Slack channel and a Command Center Teams channel, and anybody in our organization who wants to know the pulse of what’s going on in IT can check those two channels,” said Parpart. “We’ve been working diligently the last year on automating messages into those channels. Instead of our operations team having to manually type things in, we want alarms to automatically go in which brings us to our ServiceNow integration.”
“When an alarm is generated, we want ServiceNow to automatically create the incident ticket,” said Parpart. “We’re going to push into ServiceNow. We’re also looking at ServiceNow returning back. As we look at the CMDB and we look at change management and those tickets, that’s a two-way flow.”
And that’s not all the University of Chicago plans to automate. Power, cooling, access control, LDAP, electronic door locks, and video surveillance are all on the roadmap.
“Somebody opens a cabinet, we want to see ‘Ray just opened a cabinet, but he did not have a change ticket because it’s been validated in ServiceNow, so he’s got a problem,’” Parpart said.
“We’re also looking to tie into our cable management plant. That system ties to the networking system which allows us to track a MAC address anywhere in the organization. I can now track a MAC moving anywhere through the system which means I need our DCIM software to now know what the MAC of the machine is so I can find where it is on the other end.”
“What we’re trying to do as an organization is to get out of the data world and into the information world,” concludes Parpart. “We are the colo for the university. Anybody here at the university who’s got a system can put their systems in. We’ve got to communicate with all our researchers on what’s going on in their environments. If we see our central chill plant has a problem, thermo problems on equipment, or if anything starts to go awry, our DCIM is the center of our universe, but we need that tool to communicate proactively across the organization.”
Bringing It All Together
Cutting-edge data center professionals are realizing the enormous potential of driving automation in their data center operations. Whether it’s automating capacity planning to save money, automating back-office processing to save time and reduce human error, or building an entire automation architecture to turn data into information, the possibilities are endless.
Learn from the best in the industry and consider deploying second-generation DCIM software to drive automation in your data center.
For more information on how DCIM software can help you drive automation in your data center, watch the Automation Workshop webinar recording.