It might seem counterintuitive for a cloud provider to opt for on-premise backup and recovery, but Australian ERP vendor Pronto Software found it was the only way to guarantee service to its customers.
Pronto’s cloud services division auditioned infrastructure-as-a-service (IaaS) providers Amazon Web Services, Rackspace and IBM SoftLayer, but ultimately went on-prem to ensure business continuity, says Pronto’s cloud services manager Chris Dickinson.
“We’re a believer in owning our own stack. Contractually, there’s no benefit [to public cloud] and the liability is you get two cents of nothing, so it would put our business at risk by outsourcing to an IaaS provider. So we looked at how to improve our overall scalability and to act fast from a disaster perspective,” adds Dickinson.
Pronto chose upstart recovery software maker Zerto, largely for its ability to almost instantly recover to a recent point in time, reliably resuscitating VMware virtual servers and data down to the customer level. “We can deploy and react very quickly,” says Dickinson.
Pronto had Veeam and VMware but while “it would do the job and be acceptable in the market, it was not acceptable to us”.
“We wanted something to do that block-level storage,” he says, referring to speedy, incremental backups.
Fortunately, Dickinson is yet to recover from disaster under Zerto, but he says a simulation showed that there would be performance improvements of several orders of magnitude over a traditional recovery.
“From old-style disaster recovery, it would take an engineer an hour to get a customer live and operational. Now we just push a button and an engineer can bring up 20 to 30 customers in 20 to 30 minutes. That is at least 80-90 percent more effective as opposed to using database replication.
“When you have 230 customers and bring them on in less than 24 hours, that’s incredible.”
Pronto has 1,600 customers, from small manufacturers and distributors to big-box sellers, and many are moving to Pronto as a service. “[Zerto] allows me to deploy my people to do other initiatives that drive revenue,” says Dickinson.
Zerto APJ director Andrew Martin says the software is “not really a recovery; it’s a failover… If site A goes down, you should be able to bring it up on site B in minutes,” Martin says. “Zerto was born in the software-defined world. Traditionally, replication was done in the storage layer but as the world becomes virtualised – and the data centre is software-defined – replication should sit inside a virtualised hypervisor.”
The higher abstraction enables more reliable resurrection of virtual servers, he says. “In a virtualised world, applications tend to be split across virtual machines. If you don’t understand interdependencies, when you failover you have no guarantee of consistency because the order those machines must be brought up isn’t understood.”
Bang for backup bucks
Managed service and cloud providers are keen segments for Zerto. In WA, Zettagrid is emboldened to utilise Zerto for more sophisticated projects, says the infrastructure service provider’s chief technology officer Nikki Pereira.
“You can look at this product as backup and replication [but] in America, people are using it for migrations – moving from one provider to another and on-prem and off-prem. Zerto talks between hypervisors and soon AWS as well so people can start moving workloads.”
As Australian network connectivity improves, Pereira sees this approach to active use of backup data becoming more common. He says that, for example, you could have a network problem in Brisbane, but replication can kick in to bring those services back up elsewhere (such as in AWS) in four seconds, so there is no downtime or loss of services.
He also says business owners demand ‘real-time’ backup and recovery and a desire for insight or other benefits beyond basic business continuity: They want to see other features with it.”
And new markets are opening in professional services. “Lawyers, accountants and stockbrokers – people had written them off as not interested in enterprise technology but it just wasn’t available.”
Next: making backups more useful