How Leading Data Center Managers Track and Manage Parts
Data center managers often struggle to accurately track and manage their data center parts inventory. Legacy management tools like Excel spreadsheets are commonly used, but they fail to support the complexity and distributed nature of modern data center environments. They are hard to use, difficult to maintain, time-consuming, and error-prone.
Instead, cutting-edge data center managers are increasing uptime, efficiency, and productivity by leveraging Data Center Infrastructure Management (DCIM) software to manage parts. New functionality in second-generation DCIM software increases the granularity of data center asset management by enabling data center professionals to centrally manage all their hard drives, cards, memory modules, cables, and any other component, even boxes of screws.
In a recent Sunbird customer workshop, data center experts from Merck and Workday shared their real-world use cases on how they manage parts and spares with DCIM software.
Merck Moved Off Spreadsheets to Simplify Parts Management
Merck, like many organizations, used to rely on spreadsheets to track their data center parts. As Jeff Carlton, CTC Data Center Engineer/DCIM Global Data Center Engineering put it, this was “the old way of tracking our inventory at several different sites all managed by different people and different teams.”
“We typically had two sections of our spreadsheets: the assets section and the parts section,” said Carlton. “When we brought on DCIM, the assets section got taken care of with dcTrack, but the parts were still a problem and remained within the spreadsheets. This comprised of our memory, hard drives, power supplies, SFPs, and PCI cards.”
Soon, Sunbird introduced its Parts Management feature and Merck was an early adopter, eager to resolve their pain points of manually managing multiple spreadsheets of parts. They began by building their own parts library with customizable parts templates, a flexible feature that allows them to define the parts they want to track and which asset they are associated with. Merck also leveraged custom fields to track any attribute about their parts that they desired.
“With dcTrack’s Parts Management feature, we’re able to build out our parts library based on the spreadsheet data we had, which was over 300 entries of parts within this library,” said Carlton. “Many of the parts we have are bulk, so we chose to organize them as stock items to be counted. We also added two additional fields: the owner, because several of the platform teams own different parts, and the date when the part was received at the location so we can track how long it’s been here and whether or not the team still wants it after a certain period of time. While we’re managing the stock, the owners are still informed of what’s going on.”
Setting up their parts inventory was fast and easy.
“To get the rest of the data from the spreadsheet, we set up the bulk import templates with our parts data and loaded it all into dcTrack. We used the location of our inventory room and the reference field for the bucket or shelf that the part was sitting on at that time. For last-minute counts… our inventory people were able to update the tool directly.”
By configuring thresholds on part counts and enabling alerts to be notified when thresholds are violated, Merck can know exactly when they are running low on a certain part and need to resupply.
“Now, the parts inventory is managed by each of the sites by our operations teams, and any site is able to view what is or is not available within our inventory selection at multiple sites,” said Carlton. “We are able to track the usage and, as certain items are consumed, once we get to a low inventory stock, we are able to use notifications to keep up with if a particular part has gotten too low and whether or not we need to notify an appropriate team to purchase more of a certain thing like SFPs or hard drives.”
“We were able to take our list off our spreadsheet and generate the standardized library of all of our different parts models. Taking that data and applying it to the actual list, we’re generating a large inventory of over 500 parts now between two data centers. These two data centers are able to compare their inventories collectively and say, ‘We’re out of one part at one, then we can ship it to the other,” or if there are any other issues, they can negotiate back and forth.”
With a fully searchable, sortable, and exportable audit log of all parts transactions, Merck can track everything that’s happening with their parts.
“Looking at the transaction for a part, we’re able to keep up with who is consuming them for a project. It allows us to keep track of what’s going on and align part usage better than what we had seen using a spreadsheet,” said Carlton.
Workday Automated Parts Management to Eliminate Manual Data Entry
Workday is a leader in automating data center operations, and they have done many impressive things to automate parts management. Moshe Haber, Senior Dev Ops Engineer, explained how all this is possible.
“We have our own in-house development of a REST API, a generic API that every application has its own plugin to,” said Haber. “The reason why we’re using this generic REST API is because all the addons that we have (authentication, authorization, and also monitoring of all those transactions going through this API) and we can provide statistics. We’re using Sunbird’s API and also their access to the ODBC to access the tables.”
“The architecture is based on Jenkins,” said Haber. “Jenkins is a very good tool for automation, and it has a lot of plugins that allow us to provide this functionality. We have a plugin for Jira, a plugin for Slack, a plugin for GIT, and the API. The automation is based on multiple jobs that are processing data and generating the reports or alerts.”
First, Workday has daily, monthly, and quarterly reports to keep data center managers aware of the details of their parts usage and actions.
“Each report goes to each data center’s manager and the site admin responsible for the parts at any given site,” said Colm Russel, Data Center Operations Engineer. “Our daily reporting covers the location, the project number which is where we would record our Jira numbers, the assets’ serial numbers, our Workday part IDs which are the BOM IDs shared between our parts and our assets, the warranty status for the asset, the tracking number where we record our vendor case IDs, the action taken on a part, the part vendor, the part model, the part number, the field that was adjusted, what it what was changed from and changed to, who it was changed by, and the date and time it was changed on.”
Workday’s monthly parts reports are very similar to their daily reports, but they cover the entire month. A CSV file of the report is also provided which allows them to sort and filter the results for easier management.
As for their quarterly reports, “this focuses on our purchase actions where we just like to see what we are spending per quarter on any given part with any particular vendor,” said Russel. In these reports, Workday records the location, project number, reason for purchase, PO numbers, action taken, part vendor, model, and number, and the number of parts in use and in stock.
Haber explained how Workday’s daily, monthly, and quarterly reports are generated.
“Those reports are Jenkins jobs on a schedule, either daily, monthly, or quarterly. Via the API, we retrieve the tables. In the case of daily reports, we read all the transactions in the last 24 hours, we process the transactions, and we generate a report. In addition to the daily report, we also look at all those parts that were consumed and check for the ‘return from vendor.’”
Workday has also used DCIM software and automation to make it easy to understand the status of their warrantied parts.
“Part of our daily reporting features a small section which is called Consumed with No Return from Vendor (More than 10 Days),” said Russel. “We have contracts with our vendors for all our in-warranty assets. Part of our entitlement is that during break-fix, any components we may have to replace, we are entitled to receive a component back. We should receive our return part within a certain timeframe. In order to make sure that we have received that part or that the part is no longer just sitting at reception… we have implemented logic which checks the system, takes a look at a part that is in warranty, and sees what date the consume for that part was performed on. If there has been no return from vendor transaction processed through dcTrack for the same part within a 10-day period, then we receive an alert.”
Workday’s alert includes the tracking number with their vendor and the Jira number that’s associated with it. This tells them that, for example, 10 days ago, they consumed a part but have still not seen a ‘return from vendor’ action taken in the system. With this information, they can follow up on the Jira and the engineer in question to confirm whether the part has been received and ensure the proper protocol is followed.
Workday has also automated their parts deprecation process.
“Automation determines if a component can be deprecated based on its association with an installed or stored asset,” said Russel. “Both assets are all associated with a BOM ID. All parts are also associated with BOM IDs according to the server models they are compatible with and the Workday build for that particular server model. Any given part may be associated with multiple server models or multiple BOM IDs. The system will determine if these are in use anywhere in the system. If it determines that there are no live assets in the system using these BOM IDs, then the parts associated with those are considered to be deprecated. A part will only be deprecated in the event that all BOM IDs that it is associated with are no longer active in the system. It will generate a Jira to that effect, and this Jira will be assigned to each data center to confirm the status of the part in question.”
Next, Workday automates Jira comments regarding actions taken with parts.
“We have comments that are made to Jira based on the following parameters recording during ‘adjust stock’ actions which are then recorded to the parts transaction log. We have the action taken (consume, return to stock, return to vendor, etc.), we record our Jira number in our reason field, we record the asset’s serial number in the reason field, and we record the warranty status. We are able to use our quantity and we then have further automation surrounding that to comment ‘performed by,’ ‘component type,’ and ‘changed on’ date.”
Demonstrating a Jira comment, Russel continued, “We have the Jira number, the part number, the component, its description, the action taken against it, who it was performed by, the asset it was performed upon, and the date and time that was changed.”
Haber explained how Workday uses Jenkins to achieve this.
“We have Jenkins jobs running on a 15-minute schedule. Every transaction that goes into the transaction log, we retrieve the information of the transaction log, we find the associated Jira, and based on the action, we add comments in the Jira ticket. It’s done on a schedule and every action is getting a comment updated in the ticket.”
On top of that, Workday automatically generates Slack notifications.
“We then utilize Ansible automation to take a look at those Jira comments, and in the event that an engineer may ever close a Jira without performing any of their parts tracker actions on an in-warranty or out-of-warranty part, we have set alerting up,” said Russel. “The Slack notifications will go to the engineer that the Jira is assigned to and also to a general queue where our parts tracker site admins all have visibility… It’s just a reminder to make sure that you go back and complete those actions.”
Bringing It All Together
Legacy management tools like spreadsheets no longer get the job done for modern data center professionals. They require modern, elegant software that enables them to maintain uptime, increase the efficiency of capacity utilization, and improve the productivity of people better than ever before.
Parts management is an integral part of data center asset management, and without an effective parts management solution, you will not have the most accurate and complete view of your asset inventory.
Want to see how Sunbird’s world-leading DCIM software makes it easy for you to manage all your parts? Take a free test drive today.