»

Get it right for when it all goes wrong

Disaster recovery is well-understood, well-priced and rapidly becoming essential. Are local businesses investing in it, though?

Mikey Molfessis, Mimecast (Karolina Komendera) Mikey Molfessis, Mimecast (Karolina Komendera)

Disaster recovery is well-understood, well-priced and rapidly becoming essential. Are local businesses investing in it, though?

Business continuity and disaster recovery (BC and DR) are IT fundamentals, and they’re well-understood and readily available. And yet, as so many businesses have found out to their cost, the level of protection that an organisation thinks it has when something happens to cause downtime in critical systems, and the level of cover it actually has are very different things.

All too often, backups either don't exist, or can't be quickly put in place when servers start to go wrong.

Hemant Harie, Gabsten Hemant Harie, Gabsten

To find out whether South African businesses are investing in the right solutions, and how vendors and service providers can help customers get it right, we put together a roundtable of industry experts to look at the state of the market and some of the challenges that are commonly faced.

Many of our roundtable discussions focus on the importance of cloud services in today’s business world, but if there’s one set of features that is undeniably delivered best from the cloud, it’s BC and DR. That’s down to one thing.

“Connectivity is key,” said Iniel Dreyer, MD of Data Management Professionals (DMP) SA. “As long as you have connectivity, you can deliver it from anywhere in the world.”

It’s an easy proposition for customers to understand – services delivered from the cloud are by definition off-site and can be replicated in multiple datacentres, providing the peace of mind businesses need. This, said Hemant Harie, MD at Gabsten, has driven interest from small and medium-sized businesses.

Neil Edmonds, Dial-a-Nerd Neil Edmonds, Dial-a-Nerd

“The cloud has made it more convenient, but it’s also made it affordable,” Harie said. “There’s no capital expense.”

“There’s huge growth in adoption,” DMP’s Dreyer said. “Cloud adoption has been great, but now people are being more specific about what they put in the cloud, and BC and DR are key areas. The biggest challenge in South Africa is that customers are not educated as to how to do it properly. There are a lot of people selling tin in the datacentre and a piece of wire, without paying attention to what a customer actually needs. It’s up to us to do that education, otherwise people perceive it in the same way as an insurance policy, so they buy cheap.”

Customers trust us with their data, Dreyer continued, and their data is their livelihood. There’s a responsibility to be more consultative in sales than many currently are. 

One key issue, said Mikey Molfessis, cyber security specialist at Mimecast, is that customers conflate BC and DR with simple backups.

“There’s a perception that if you move a service to the cloud, and it’s backed up in the cloud, you automatically have DR and cyber resilience,” Molfessis said. “There’s a big difference between high availability, backup and a DR strategy. People don’t think that ransomware exists in the cloud, for example, or that they won’t get attacked. But they don’t know what they don’t know.”

Complex relationship 

Steve Briggs, Seacom Steve Briggs, Seacom

The implications of service level agreements (SLAs) aren’t always fully understood, for example. Molfessis cited one financial customer that committed to respond to its investor queries within 30 days, and felt confident that its three-day SLA on email services would allow it to meet its obligations. “What happens if your servers go down on day 28 of receiving that email?” Molfessis asked.

In the same way, said Johann Els, open source manager at SUSE South Africa, it’s not enough to focus on key applications either. “To get to the right DR plan, you need to think about the design of what you define as high availability,” he said. “What services do you need up, and what do they need to function with? The relationship between services is very complex.”

Johann Els, SUSE South Africa Johann Els, SUSE South Africa
Customers might have petabytes of database information backed up, Els continued, but what does that mean when there’s a problem?

“If a server is dead, you have to think about how you recover that amount of data and what’s the process for moving it around?”of people selling tin in the datacentre and a piece of wire, without paying attention to what a customer actually needs. It’s up to us to do that education, otherwise people perceive it in the same way as an insurance policy, so they buy cheap.”

Customers trust us with their data, Dreyer continued, and their data is their livelihood. There’s a responsibility to be more consultative in sales than many currently are.

Driaan Odendaal, Arcserve Driaan Odendaal, Arcserve

“It’s easy to put stuff in the cloud,” added Driaan Odendaal, technical lead, Arcserve Southern Africa. “It’s harder to get it back out when you need it.”

“People don’t think about getting it back into the live environment,” agreed Neil Edmonds, technical manager at Dial-a-Nerd.

DR to data audit

Developing a true BC and DR plan is a process that involves mapping out services and understanding the relative value of the data they own – it’s a key step towards true digital transformation, in other words.

“It comes down to prioritising applications,” said Mimecast’s Molfessis. “I may need to have email available all the time, but I might have a source code server, which is fine if that's out for a day.”

“Understanding what data you have in the first place means knowing if it’s worthwhile paying a premium to keep it available,” said

Commvault’s channel manager Gerhard Fourie. “You also have to consider that there’s a risk to big uptime services with replication, too – you could be replicating the corruption that caused the problem you’re now facing.”

SUSE’s Els said he’s a fan of the 'chaos monkey’ approach popularised by Netflix, and simulating random failures to see what effect they have and how long they take to consider.

“It’s about figuring out where my availability zones are,” he said. “What happens when any part of a zone goes down? Can I really recover? Sometimes I take business and IT into the datacentre and I say, ‘What's this?’ Can I pull out this cable?’ And often they don't have an answer as to what its effect will be.”

Iniel Dreyer, Data Management Professionals Iniel Dreyer, Data Management Professionals
The participants agreed that IT systems tend to develop over time, and are poorly documented. Often, the person responsible for a particular key server has left the company and no one else is fully capable of restoring it. Commvault’s Fourie told a story about one customer who was struggling to physically locate a machine that was attached to its network.

“Eventually, they found it behind a drywall,” he said. “It had been built over the last time the office was remodelled.”

A new relationship

There’s still the challenge that many organisations are only willing to engage in serious conversations about BC and DR best practices when it’s already too late.

“It’s often only after the first attack that you have a conversation,” said Steve Briggs, chief commercial officer at SEACOM. “Or when you realise that just because you have an SLA at the datacentre, that doesn’t cover you for the guy who dug up the networking cables at the traffic light down the road.”

“It’s a cost thing,” said DMP’s Dreyer, and part of the negative perception is around the way DR and BC is currently sold. “People are very price-conscious and you end up leaving out resiliency from a solution to keep the cost down; then you try to upsell it as an add-on and it becomes a grudge purchase. It should be built into everything; we shouldn’t be selling solutions that aren’t resilient.”

Dial-a-Nerd’s Edmonds agreed.

“One of the things I've noticed is that we’re often talking to the finance people,” he said, “who are looking at how this fits into the budget.”

Gerhard Fourie, Commvault Gerhard Fourie, Commvault

DR and BC is really a business conversation, however. Large organisations often look at a cost of downtime calculation and feel that they can carry that, added Mimcast’s Molfessis, without thinking about what the business and reputational costs are beyond the financial numbers.

Still, as DMP’s Dreyer pointed out, if the goal is to build trust and move to a consultative model, those involved with disaster recovery have an advantage over other service providers.

“You make friends at 2am on a Saturday morning when they phone in a panic and you help them out,” he said. “That’s when you earn trust and they will listen to you going forward.”