Thiess cuts failover to disaster recovery time by 83 per cent
- 15 July, 2013 11:18
Chris Montgomery, infrastructure manager for Thiess ICT
The 2011 Brisbane floods were a wakeup call for Thiess: The engineering services firm's legacy storage system required a complete shutdown to migrate data, which had to be done within 24 hours in order to minimise any interruption to its business.
The combination of having data centres sited in flood-prone Queensland and data growth of 90GB a day, meant that Thiess' storage and backup/recovery environment were not ideal in high-pressure situations, said Chris Montgomery, infrastructure manager for Thiess ICT.
Thiess has two data centres in the state: One at its own site in South Bank, and the other at Eight Miles Plains, a hosted facility which it rents from Interactive but manages it itself.
“If anything happened with any system and we lost data in any way, we wanted to be able to recover it quickly,” Montgomery said. “We didn’t want to run into situations where there was conflicts between backups and restores; we wanted to be able to do the restores at the same time and continue to backup as per normal.”
In the aftermath of the floods, Thiess followed advice from Bridge Point and deployed NetApp MetroCluster and CommVault's Simpana solution, cutting its failover to disaster recovery time by 83 per cent.
“We do a complete disaster recovery test at least annually as a minimum where we literally failover to DR everything between data centres,” Montgomery said.
“It’s not a simulated test it’s a real test. In the 2011 Brisbane floods we came within 24 hours but in our test since then we have got it down to four hours.”
Having replaced 80 per cent of its storage infrastructure, Thiess was able to consolidate four different types of storage systems down to one. Montgomery said although the previous systems were all provided by one company, finding the resources and skills to operate the several management interfaces was costly and time consuming.
“We were able to cost-effectively train multiple people to understand the system and not have a dedicated storage administrator. So we previously had somebody that would have been not full time but a person that was largely dedicated to working on storage. That is now something that’s become part of our standard operations without any sort of staffing burden.”
Previously Thiess had to shut down all of its applications in order to do a complete failover to disaster recovery – now Montgomery can do it in real time without interruption. With the majority of systems now virtualized, he is able to fail the applications over with ESX from one server and data centre to another in real time.
Montgomery said the only challenge he has had so far with the new solution is adjusting to the NetApp architecture. However, this has involved a relatively shallow learning curve compared to the training his staff would have to go through with the previous systems.
Thiess is also embarking on a project to use Riverbed for its wide area file system, Montgomery said. The project is currently in deployment stage and will be complete by the end of October this year.
“For people out on site who are accessing services delivered from Brisbane, it optimises the traffic across the network to get the user a better experience – the time to open and save files is quicker.”
Follow Rebecca Merrett on Twitter: @Rebecca_Merrett