A large consumer products company had their data center in the basement of their headquarters’ building. When the basement flooded due to a pipe that burst and brought down their systems, they decided that the current location was too risky.
It was time for a data center migration. Since the equipment was in the building where most employees worked, however, everyone was used to applications responding rapidly. After the move to a location 30 miles away, the company needed to ensure performance would be as good or better.
Bill, the IT director, decided to make the move when much of their hardware was coming off lease. This timing meant they would not have to move all their old equipment to the new location. Also, it gave them an opportunity to buy new hardware and improve performance.
They took the following steps. You can duplicate this approach to make sure your applications stay responsive as you migrate data.
Create a Profile
Before getting started on their data center migration, they profiled its capacity, IOPS, latency, throughput and other vital statistics. These measures served as the baseline for building out their new data center. To create the profile, they used their infrastructure performance management solution.
Size the Network Pipe Right
The five hundred or so people in their building were used to getting rapid responses on their computers from the data center in their basement. Bill was concerned that when they picked up the data center and moved it thirty miles away, communication would slow down over the Internet. To ensure good performance, they needed to size their network pipe correctly. This issue is, by the way, one that many IT leaders overlook.
The bandwidth issue is one you can understand more easily if you think of your mobile phone. When you Google a question in the office where your phone is on the network, you can get an answer more rapidly than you can at the grocery store. If you’re out shopping, communications come over your cell connection, and the reduced network bandwidth impairs performance.
To address this concern, they reviewed the level of network activity coming in and out of their data center which enabled them to specify the right bandwidth.
Optimize Performance and Costs
While Bill was interested in ensuring application performance, he needed to do so as cost-effectively as possible. When he looked at their IT infrastructure’s profile, he saw that a lot of their servers were either unused or underused. He decided to downsize from 100 to 70 servers and increase utilization levels. Since some of their servers were coming off lease, they bought the replacement equipment and had it shipped directly to the new data center.
Migrate Applications One by One
Bill wanted to take a conservative approach to ensure applications ran well throughout the transition. So he decided to migrate the applications one at a time. The non-production, less critical applications lead the way and the mission critical ones were last in line. After all, if there was a problem, it was better to affect the developers than, for example, the customer relationship management system or website.
It was a learn-as-you-go process. When the IT department moved an application, they monitored its performance to see if it was performing better, worse or the same in its new home. If any application failed to show increased performance, they dug into the metrics, found any issues and resolved them. By the time it came to the mission-critical applications, they knew how to migrate applications successfully without missing a heartbeat.
Counter Perceptions with Facts
Even though they did everything right, the IT department fielded a few complaints. After all, if employees know a data center migration is underway, and someone has a bad day, it’s easy for the IT department to become the scapegoat.
Because they had data from monitoring the IT infrastructure before, during, and after the migration, they were able to put objections to bed rapidly. The system administrators looked at their infrastructure performance management tool and said, for example, “You’ve been running this job every week on Mondays at 11 a.m., and it’s been taking five minutes. Now, on this new equipment, the job lasts four and a half minutes. While you may think it’s running slower, it’s actually thirty seconds faster.”
Toot Your Horn
After everything had gone smoothly, it was not the end of the process. Bill’s team continued to monitor the new environment. Looking at the key performance metrics, Bill could say to company leaders, “My team moved the data center to a new environment, and we shrunk the software, hardware, power and computer room costs by around 25 percent. And, by the way, everything is running faster.” Then he provided the proof: colorful charts from their infrastructure performance management solution.
So if you want to handle a data migration well and minimize risk to applications, monitor performance before, during and after the migration. Use the data to determine your specifications for your new data center, optimizing performance and costs. Gradually move the applications, putting the least important ones at the front of the queue. Use your monitoring tool to detect any issues and correct them as you go. By the time you get to the most critical applications, you’ll know exactly what to do to make the transition trouble-free. Once you’ve completed the migration, continue to monitor performance and use the data you collect to counter any objections and, of course, to make upper management aware of the value you provided.