- KEY TAKEAWAYS
- The global VDI market is forecasted to grow by more than 27 percent from 2016-2020. Higher education must adapt to a rapidly changing market to keep pace with the demand in digital learning via virtualized platforms.
- While numerous institutions have invested in virtualized servers, many are beginning to virtualize desktops to save money and simplify access to their users. IT leaders are also using virtual desktops within shared computer labs to give students access on their desktops and personal devices. Virtualized desktops are also replacing administrative desktop PCs to save money and improve security.
- Tintri VMStore enhances visibility into the VM-storage environment for fast deployment as well as accurate and early capacity planning. This approach lowers costs and dramatically improves speeds and reduces the complexity of provisioning, deploying, and maintaining technology footprints.
Tintri Scale-Out and Tintri Analytics brings VM-awareness to a new layer of VM management. We’ve already pioneered VM-aware storage arrays. Now, it’s time for more.
University of California Irvine (UCI) is a public research university and one of the 10 general campuses in the massive UC system. UCI currently enrolls more than 30,000 students, and employs more than 1,100 faculty members and 9,000 staff.
UCI relied on storage arrays from one of the traditional, legacy vendors. They needed faster-performing alternative to their disk-based solution that could keep up with demands along with a way to measure performance and diagnose system performance problems. Even with the vendor’s monitoring software they couldn’t tell which VMs were causing the slowdowns.
After implementing Tintri’s VMstore, UCI can detect performance issues from data off the vCenter ESX hosts, without any guesswork. Tintri reports on every VM as well as at the VMDK level. Tintri also uses ‘thin on thin’ provisioning of VMDKs on NFS. UCI can also add encryption at any time, while other solutions require additional self-encrypting disks that are available only at the time of purchase. UCI can now add encryption when they need it.
Tintri provides the insights UCI needs to quickly diagnose troublesome VMs.
“We came in one morning and noticed that one of our VMs had ‘gone berserk’ overnight,” said John Ward, UCI IT enterprise architect. “We have a 10Gb connection and this particular VM was driving 3Gb of the network bandwidth. With just three clicks, Tintri could tell us exactly which VM was generating all of the load. We were able to fix the issue quickly with that knowledge. But the impressive thing was despite all of the IOPS and bandwidth this VM was consuming, everything else on the Tintri was still performing very well with under two milliseconds of latency.”
Washington State University (WSU) is a nationally recognized public research university with more than 26,000 students. The College of Engineering and Architecture (CEA) at WSU wanted to virtualize student desktops to improve performance, reduce costs and improve security. Initial attempts to deploy a VDI solution were unsuccessful.
The university’s existing storage systems had become a bottleneck and were difficult to manage. IT administrators could get performance and capacity utilization at the datastore level on existing storage systems, but not at the individual VM and vDisk level. Extracting performance metrics from existing systems was cumbersome and rarely helped troubleshoot issues in real-time. Hidden software licensing costs were also a problem, as WSU had to pay for every little data management feature, on a feature-by-feature basis. There was no easy way to quantify the costs up-front.
“Since implementing Tintri VMStore for VDI, we don’t have performance bottlenecks,” says Ryan Makamson, WSU systems administrator. “The CEA at WSU has deployed hundreds of virtual desktops on a single Tintri system, resulting in substantial operational savings. There are no hidden licensing costs, which helps in this tight higher- education budget environment. Tintri’s GUI provides granular performance and capacity metrics, which simplify administration – we would have certainly benefited from this on our previous arrays.”