Unpacking Murray & Roberts’ turbulent journey to the cloud and back again

It was only a few years ago when ‘digital transformation’ was on every CIO’s agenda, and businesses started to understand how cloud could deliver real value. They stopped asking whether they should move to the cloud and started asking what they needed to do to get there.

This was the case for Murray & Roberts’ CIO Hilton Currie in 2016, when the cloud services market in South Africa was booming. Already worth $140 million and growing rapidly, it was around this time when the engineering and mining contractor undertook its migration to the cloud. But things didn’t go according to plan — a story not unique to South Africa or this industry. So Currie made the difficult decision to repatriate Murray & Roberts’ IT stack, one that required selling the business on reversing an IT strategy Currie had previously sold as M&R’s future.

Currie recently spoke to CIO.com about the mining company’s bumpy cloud journey, what motivated them to move and, ultimately, what drove them to change back. 

CIO.com: What motivated Murray & Roberts to embark on its original cloud journey?

Hilton Currie: Up until 2015, we were 100% on-premises and we had no major issues. I think the first red flag was the age of the equipment in our primary and secondary data centers. We pretty much ran on a major production environment, and then we had a hot standby disaster recovery environment, which was far out of support and had come to end of life. This brings its own risks to the table. Our production environment still had a bit of life in it, but not much and decisions had to be made.

Around this time, we were in discussions with our outsource partner to outsource our IT support and they proposed that they could make cloud affordable for us. Honestly, in 2016, cloud wasn’t really affordable. If we were willing to look at outsourcing the whole lot to them — all the way from server application right down to the technician support level — and adopt their managed public cloud, they were confident they could make it work for us. If we looked at the cost of new infrastructure we needed to renew, it actually looked very attractive financially. So we opted to go ahead.

Once you signed on, how then did the migration progress?

We started with a serious migration from the first quarter of 2017 because a lot of our equipment was end of life, so it was an all-or-nothing approach. We used a locally hosted cloud vendor that was connected to our outsource partner, but also with affiliation to a big global company. It took a bit longer than anticipated but by November 2017, we were almost fully across and functioning. We had everything from our big ERP systems to smaller, bespoke systems running in the cloud.

We had third-party independent consultants come in to analyze certain systems and licensing. But in early 2018, the first major hiccup hit us as the independent party who did the audit missed the terms of use on some of our licensing. For example, Microsoft has quite heavy restrictions on using perpetual license software in the cloud, specifically SQL Server. Some licenses aren’t valid if you don’t own the rights to the lower-level equipment. Unfortunately, our licensing expert missed that. To adjust this, we’d have to move from perpetual licenses to subscription licenses, which would have been a grudge purchase because some systems lag in terms of the versions they’re certified to run on. We would have had to purchase a current version of SQL Server and then downgrade it like four or five versions because that’s the version our ERP and other systems run on. This would have been hellishly expensive, so we brought all services that were impacted by licensing restrictions back on-prem and pulled all our SQL servers back, which was a costly exercise because we had to purchase new equipment, and the rest of the applications ran in the cloud. We ended up with a split or hybrid setup, which brought some challenges and became quite a nightmare to manage. We did eventually get it working and ran this way for about six to eight months. During that time, there was a buzz around cloud and I think the outsourced vendor was getting a few new clients on their managed cloud platform, which started taking a toll on us because it wasn’t long after that they started imposing rate limitations.

What prompted you to realize the shift to the cloud wasn’t working, and how hard was it to decide to move back?

Historically, we could get full performance out of the cloud platform. There were no restrictions and then suddenly they started imposing limitations, with an additional charge if we wanted more. Suddenly, the commercial model fell to pieces. We tried to make do with the limitations but it got to a point where it just wasn’t working. We were only fully cloud for about a year and a half and about a year in the business was brought to its knees. Applications started failing, email and phones didn’t work and our ERP became unusable. In some of the worst-case scenarios, it took our finance teams up to 15 minutes to open Excel files. We complained and they lifted some of the rate limitations while we made alternative arrangements.

Around March 2019, we decided to move back on-premises and by mid-2020 we were fully on-premises again. For me, the decision was crystal clear. At a point, it felt like I was walking around the building with a target on my back because this affected everyone and there was an air that our IT was falling to pieces.

In Murray & Roberts, IT reports to the financial director of the group, so I sat down with him to explain that it’s not all bad. I built a detailed roadmap highlighting that we were in a bad space, but outlined that getting out of this situation was possible. It came down to the numbers in terms of spend on new kit. I showed that it would cost less over three years moving to new kit than it would staying where we were. It was a no-brainer for him to accept but it was a difficult conversation. We sold them the cloud journey back in 2016 and they backed us and jumped on board. Then, a year and half later, we wanted to jump ship. But I think the results of moving back to a private cloud spoke for themselves.

So what systems are in place now in M&R’s on-premises data center? Have the issues you identified been resolved?

Instead of keeping things as is when we moved back on-prem, we did a refresh by making a list of all our systems to get a better view on how important they were to the business and where they sat. As part of this rationalization process, a couple of the systems were upgraded since we had a chance to rebuild from scratch, so we took advantage of that and got things running the way we wanted. We did a lot of consolidation as well. When we went to cloud, at one point we had over 300 servers and the end goal when we moved back on-prem was to get this down to about 180.

Given how the cloud market has changed, would you consider another cloud migration?

We’re not against cloud. We understand it can add value and it has a place. In fact, we’re starting a complete Office365 migration. But we’ll only lift certain systems and we’re taking a more selective approach. Cloud is a big buzzword, but you need to ask what it promises and what it’s going to give your business in terms of value. If you’re going to cloud for commercial reasons, it’s a big mistake because it’s not cheaper. And if you’re going for performance reasons, it’s an even bigger mistake and there are many reasons for this.

In South Africa specifically, there are a lot of issues of bandwidth and throughput to international vendors because stuff still sits in Europe or in the Americas. With the kind of flexibility that virtual environments can offer on a private cloud, do you really need a public cloud if you don’t need to be agile or scale drastically? We found that a well managed, fully redundant virtual environment, hosted in the private cloud on our own kit, in a tier-four data center was the ideal scenario for us. We’ve been running this way since mid-2020 and have not looked back

Any learnings from this experience that you’d like to share with other CIOs?

Looking back, I don’t think we made a bad decision. The biggest learning is about focusing on the big picture. Make sure you understand your long-term roadmap very clearly so there are no surprises. Often people will be blinded by the commercials, but be very careful about licensing and terms of use because many vendors have restrictions in place. Do your due diligence. All vendors write clauses into their contracts that it’s subject to change over time. But make sure you’ve got a backup or a rollback plan because it’s very difficult to bridge the gap when the company is on its knees and you need to buy equipment and do a full migration. I would also never recommend any full lift-and-shift approach. There are just too many variables.

What advice would you give aspirant CIOs, having gone through this? One of the aspects that’s lacking in many CIOs is the interaction with the business. The CIO role is not just an IT role. Gaining an understanding of your business and building relationships with important stakeholders is key to a successful CIO career. Although you’re looking at governance, you’re looking at compliance, processes and things like that. If you’re not aligned with what the business requires, you’re sitting in no man’s land because there’s a mismatch between what IT offers and what the business needs. When you’re looking at technology for all bells and whistles, you’re missing the point. It should be about adopting the right technology to boost productivity and to facilitate how the business operates.