Des Moines University is one of the top 20 largest medical schools in the U.S. and the second oldest osteopathic college in the country. As a senior infrastructure engineer, I am part of the team that designs and maintains the IT infrastructure that supports thousands of students, patients, staff, doctors and nurses. Like any good infrastructure engineer, I am entrusted with making sure that our mission critical infrastructure is operating at peak performance, 24×7.
In 2006, DMU launched a long-range strategy to virtualize our infrastructure. Today, we’re 99% there and virtualization has empowered us to support an ever growing service catalog without adding additional staff. Today we operate in an almost entirely digital world; health records, student applications, course work, lecture videos are all handled electronically. As more of our services have become dependent on our IT infrastructure, performance and reliability is imperative in our environment.
Our existing storage, backup and disaster recovery solution, a combination of NetApp FAS3060 and Symantec Backup Exec, was the “huge solution that couldn’t.” It didn’t have the horsepower to meet our performance needs or the ability to add capacity without significant cost increases. As our dataset grew flash cache performance declined and we experienced performance problems as our active data grew. To increase capacity would mean bringing in the forklifts for an upgrade. But we had bigger issues than spindles and platters. Our existing solution couldn’t support tiered storage for balancing performance and cost, and it couldn’t meet our eight-hour window for nightly system-wide backups. We needed more bang for fewer bucks.
Our answer was to implement an EMC Next Generation VNX unified storage array in each of our two data centers. The hybrid storage array enabled us to benefit from a small amount of flash storage to meet our extreme performance needs, SAS disks for performance and near-line SAS for capacity. Less than one percent of overall capacity is flash, yet it manages 70-80 percent of overall I/O reducing latency from 10-15 milliseconds to 2-3 milliseconds in our Exchange environment. Automated tiering, which automatically stores data on the most appropriate drive type, allows us to add economical large capacity drives for growth or high performance disks for additional IO as needed. For backup, we use EMC’s Avamar software for our 32-bit file shares. Our virtual machine images and 64-bit file shares are backed up with EMC’s Data Domain technology.
- Here’s our full configuration:
- Two VNX 5400 Next Generation unified storage systems
- Avamar and Data Domain backup and recovery systems
- EMC FAST Suite
- EMC RecoverPoint® replication and data protection
So…does it really work? In a word: yes!
- The entire infrastructure is backed up every night within the 8-hour backup window. Most of our virtual machines backup in less than 10 minutes.
- Data Domain currently reports a 98.5% reduction in data from both client-side deduplication, and data compression
- Automated Tiering promotes hot blocks into high performance drives and moves bulk data to economical high capacity drives.
- Trouble ticket generation has been reduced from two per month with NetApp, to two tickets since implementing the VNX in the fourth quarter of 2013.
Of all the vendors we talked to, only EMC worked with us to run a performance assessment across our entire environment before presenting a solution. During the assessment they analyzed our IO workload, capacity trending and free space. Because EMC took the time to truly understand our current and future needs, we were confident we had the right solution. The whole experience, from first conversation to implementation, allowed us to throw out our old playbook and up our storage game.