A leading UK vegetable wholesaler has reduced its network bandwidth needs by 50% and cut recovery point objective...
(RPO) on ERP data to 15 minutes from 24 hours after deploying Silver Peak WAN optimisation.
Produce World is a grower, importer and supplier of vegetables to wholesale markets. It has one main datacentre in Peterborough plus a disaster recovery site in London and several other remote sites countrywide. The organisation’s key app is Microsoft Dynamics NAV ERP, which handles stock control and business intelligence and is served out to remote sites via Citrix desktops.
The firm’s IT team realised it needed to do something about its WAN after implementing the Microsoft ERP system and it was trying to achieve replication between sites, said infrastructure manager Richard Billington.
“We wanted to take advantage of SQL replication for Dynamics as well as replicate VMware virtual machines (VMs) and to use NetApp SnapMirror,” he said. “We had a 40Mbps IP-VPN link, but it just wasn’t keeping up. We would make changes to VMs and it would take whole weekends to catch up and slow traffic all week.”
Billington said that if Produce World wanted to deploy a new VM and for it to be copied across to the DR site, the system would struggle. “We had to be careful. We couldn’t do too much in one week and would have to stagger the work we did.”
More on WAN optimisation
So Billington’s team examined ways to improve the use of WAN bandwidth and looked at Silver Peak and Riverbed WAN optimisation tools.
WAN optimisation products reduce the load on existing bandwidth by combining data deduplication, caching and optimisation of network protocols and application traffic. WAN optimisation is often deployed to help with replication of data between sites to effect disaster recovery provision.
Eventually, Produce World settled on Silver Peak VX WAN optimisation software, which runs as a virtual appliance on VMware virtual machines at both the main datacentre and the London disaster recovery site.
Billington rejected Riverbed products on grounds of cost. “We kept being told it would cost extra in licence fees to add extra functions,” he said.
The Silver Peak deployment has allowed the company to run its operations on a 20Mbps link instead of 40Mbps, removed the need to spend time managing replication traffic and cut RPO from 24 hours to 15 minutes, said Billington.
“We used to have to manage replication,” he said. “We would limit and manage the bandwidth taken by SQL and SnapMirror so that they didn’t grind to a halt. Now they’re all set to max and we don’t have to worry about VM management either.”
Also, previous difficulties with network bandwidth had meant replication traffic had lagged well behind real-time operations and so the best RPO it could achieve was about 24 hours. Now Produce World can reinstate data from as little as 15 minutes before an issue arising.