As Syracuse University (SU) consolidated approximately 20 small data centers (one per business and academic unit) and built a new modern data center with a private cloud infrastructure, the challenge of providing enterprise-scale backup and disaster recovery remained. SU needed a unified virtual machine (VM) backup and recovery solution to enable 24x7x365 availability among its thousands of VMs and scale to support its explosive data growth. Within a two-year period, the university experienced a data explosion as its data grew from 345TBs to 1.34PBs—an increase of 386%.
In 2009, SU’s traditional backup tool simply could not scale and was therefore unable to back up all VMs daily. SU could only back up its most critical VMs—those running Microsoft Exchange and file servers. Backup was limited to once or twice a week because the critical backup jobs alone took 48 hours. The traditional backup tool was also prone to failure: One out of every 10 backup jobs was unsuccessful, making recovery that much more challenging.
“I can’t imagine backing up Blackboard using our previous traditional methodology, much less recovering backups during an outage,” said Josh Slade, Virtual and Backup Environment Lead at SU. “If that had been the case, Blackboard could have been down for hours, and that would have been unacceptable.”
SU’s 22,000 students, faculty and staff require 24x7x365 access to files, databases and applications. SU’s modern data center, which is 99% virtualized and delivers infrastructure-as-a-service (IaaS) to the university’s business and academic units, had specific availability requirements for everything within its vast infrastructure.
“Our IaaS offering focuses on providing networking and computing resources,” Slade explained. “Schools, colleges and business units drop VMs into our cloud with the expectation that we will provide data protection. We needed a backup solution that leveraged changed block tracking (CBT) to reduce the amount of data undergoing backup, supported Microsoft Volume Shadow Copy Service (VSS) to ensure successful VM recovery and was scalable to grow with us.”
SU deployed Veeam® Backup & Replication™ in 2009. Years later, data center administrators are still impressed by Veeam’s capabilities, particularly its scalability.
“Veeam is the most mature product available for VM backup,” said Peter Pizzimenti, IT Analyst at SU. “We’ve grown with Veeam, and Veeam has grown with us. An infrastructure our size can’t back up and recover using a typical backup interface. Veeam’s interface is extremely agile and easy to use.”
In addition to the user interface, Veeam Backup & Replication provides a Windows PowerShell extension, enabling SU to create a set of scripts to automate backup and recovery of its 2,300 VMs within its VMware vSphere environment. During a recent disaster recovery (DR) test, 1,500 VMs were successfully recovered within 24 hours. Based on these results, Slade and Pizzimenti are confident the entire data center can be restored within a similar timeframe.
"When we looked into the backup and DR market, Veeam was the only product that met our enterprise needs,” Slade said. “The command line options within Veeam are critical to our successful DR implementation.”
“We’re purists when it comes to Veeam,” Pizzimenti added. “We focus on backup, recovery and migrating VMs from one vCenter to another. We do use Veeam replication, but not in the traditional sense for DR. As part of the Blackboard upgrade process we use Veeam replication within our data centers to reduce the window of system unavailability to roughly 10 minutes.”
For DR, SU utilizes nightly incremental Veeam backups and sends them to virtual repositories in a secondary data center. Currently, the virtual repositories hold 2PBs of data and 30 retention points. As part of the DR plan SU has created a repository of custom scripts to restore every VM from the Veeam repositories to waiting storage and hosts in the secondary data center.
Like many enterprise organizations, SU relies on Veeam’s fast and efficient block-level incremental backup for multiple recovery points. Veeam employs VMware vSphere CBT to know what data blocks have changed since the previous backup job. Veeam also uses application-aware image processing to create a transactionally consistent backup of a VM running VSS-aware applications such as Microsoft Active Directory, Exchange, SharePoint and SQL Server. Because Veeam is VSS-aware, backed up VMs and applications can be reliably recovered.
“We’ve used Veeam for so long that we assume backup and recovery will be successful, and it is,” Pizzimenti said. “There’s nothing else on the market that we were able to find that could work in an infrastructure our size with the approach we have taken except Veeam.”
Scalability and flexibility to support explosive data growth
Using Windows Powershell to script Veeam Backup & Replication, SU was able to automate backup and recovery of its 2,300 VMware vSphere VMs. SU continues to experience a data explosion and expects its business, academic and research units to provision an additional 600 VMs and 500TBs of data by the middle of 2016.
High availability for DR
With campuses around the world, members of SU’s community must have 24x7x365 access to data. During a DR test, 1,500 VMs were successfully recovered within 24 hours. Based on these results, SU is confident that all 2,300 VMs can be restored within a similar timeframe.
98% faster backup
SU’s modern data center required backup and recovery designed specifically for virtualization. Before Veeam, SU could only back up its most critical VMs—those running Microsoft Exchange and file servers—once or twice weekly because those backup jobs took 48 hours. Veeam backs up those VMs in one hour.