Money In, Money Out: Building a Private Cloud to Double Business While Cutting Costs By 50%
Because I have very high expectations, I start any project expecting vendors to fail. But on a recent project, one vendor proved me wrong—they not only met but exceeded my expectations, helping my company reach our goal of doubling business year over year.
I am the Network Director at full-service carrier and telecommunications infrastructure wholesaler Devoli. Based in Auckland, Devoli provides our customers in New Zealand and Australia with world-class network, carrier, and wholesale services. We pride ourselves on leading with automation, which allows us to be one of the region’s most flexible and creative service providers.
And we are growing. I was brought to Devoli in 2019 to prepare the company for an ambitious growth plan: to double in size yearly. But first, we needed to improve the reliability of our entire network.
The Limitations of Public Cloud
Three years ago, we had three core platforms that supported all our voice applications and network management systems: one across Australia and two in New Zealand. The equipment had reached end of life—some of that equipment was eight years old—and it wasn’t carrier grade, either. Our servers were primarily Dell, and many of our monitoring and backend systems sat in AWS.
We run telco applications that require very high bandwidth and found that AWS couldn’t keep up with the necessary processing power. On top of that, the voice network makes lots and lots of small processing transactions, which you pay quite handsomely for with AWS. It’s significantly—significantly—more expensive, which prevents AWS from being a viable long-term strategy for us for core network and applications.
Because of our business, we already have racks of equipment consuming power. There was no benefit to us being in anybody else’s cloud. It was time to consider creating our own.
The Solution and Vendor Who Exceed Even My Expectations
One of our most critical requirements was the ability to continue supporting our customers seamlessly, even if we lost multiple sites. Our new solution needed to accommodate a complete failure across Australia, New Zealand, and even between the two countries. I also wanted to agree on a full five-year engagement and a partnership with end-to-end service. That was the only way I could sell it to the business as a true competitor to AWS rather than just a purchase of additional equipment.
I hold the purse strings for our infrastructure build and always have a fixed budget. This project was no different, as it was a significant investment for Devoli. I would have to sell any solution I chose to the Devoli board, which demanded strong outcomes after 12 months.
I took these requirements to Ingram Micro to get their suggestions. I’ve worked with Ingram for about 15 years on various projects at multiple companies, and they’ve always been a great partner. Initially, we discussed a different system that would augment our pre-existing infrastructure. Then, they asked if I had considered HPE and the HPE Alletra dHCI, which comprises HPE Storage and HPE ProLiant Servers. It looked like what we needed in terms of streamlined infrastructure and optimized performance, but would it fit our price tag? Ingram assured me that it did, and I was sold.
As a bonus, this solution had never been deployed in New Zealand or even APAC, and for me, there’s always something about being the first to make a mark.
There’s a lot of drive to go to someone else’s cloud but not as much to create your own. Because of that, this engagement was exciting and different from most projects. We gained direct access to the HPE team, including their engineers. And as soon as our Devoli engineers started talking with the HPE engineers, the process became much more proactive.
The delivery was phenomenal, especially considering it happened during the heyday of COVID lockdowns. One of the HPE engineers was also responsible for the installation, and they trained my team to perform other installations should we want any in the future. That action demonstrated that they weren’t just trying to make money off of us, but to ensure a quality install and ongoing education. That’s somewhat unusual in my experience with vendors.
Despite the challenges of masking and keeping engineers in bubbles, HPE completed the project ahead of schedule. Then we were off to the races. Once we deployed, we migrated 68 of 74 applications in one day, and nothing went wrong.
In Building a Platform for the Next Five Years, We’ve Cut Cost and Time by Half
We have two sides to the business. My teams—which handle network infrastructure and security—now run entirely on the private cloud, except for a few monitoring and alert systems in case our platform fails. The software side, where the majority of our business is all about automation, is still on AWS due to flexibility for the development team. However, over the next three to six months we intend to migrate as many applications as possible to the private cloud.
We have leveraged the HPE Alletra dHCI to build a platform for the next five years. Despite reducing the number of platforms from three to one to serve both countries, we’ve improved reliability. We’ve also reduced costs by 50% by moving the vast majority of our applications to the private cloud. Pulling applications from AWS provided some immediate cost relief, and those savings will cover the cost of the solution. The deal is essentially self-funding within five years while meeting our primary goals of increasing capacity and reliability.
There are other benefits as well, particularly in improved speed: by my estimates, we’ve seen a 50% improvement in deploying support applications, a 51–75% improvement in provisioning new workloads, and an improvement of more than 75% in deploying new compute and storage resources. We’ve seen up to a 25% improvement in lifecycle management tasks, including upgrades, and we spend 51–75% less time troubleshooting VMs and infrastructure issues. And the most important stat of all: maintaining 100% uptime.
One of my biggest surprises has been the proactive service we receive from HPE. Instead of waiting for us to flag an issue, we will get an email from the HPE team to say, “One of your discs isn’t operating as it should. There will be someone there to fix it this afternoon.” It feels strange to note that because no one wants to admit things go wrong, but things do go wrong in IT, and I’ve never experienced this kind of response from a vendor.
The alerts from HPE InfoSight are beneficial to a small team such as ours. It might take us days to identify a small issue, but HPE InfoSight proactively identifies problems and solutions, eliminating the need for calculations. My infrastructure architect logs into the platform weekly to perform the health checks, and it’s as simple as that.
Capacity planning isn’t something that’s easily automated. You always put out fires before you deal with strategy, and sometimes it’s hard to make space to plan at all. HPE InfoSight fits nicely into how our business thinks and builds planning into the background. It makes the team’s routines significantly faster and is an excellent addition to our existing team.
Improved Speed and Resiliency Help Us Punch Above Our Weight
All of this adds up to improved efficiency and the speed at which we work. Speed is our point of difference. Everyone has the same network. Some might make it look shinier, but we all deliver the same product. Devoli’s differentiator is our ability to deliver any broadband voice service within a few hours—not weeks or months. We have some large customers who consume a lot of voice traffic. Now we have the enhanced reliability to move faster, which has garnered us even more business.
Devoli punches well above its weight in terms of how we compete against bigger players here in New Zealand. Some organisations would try to compete by cutting corners, whether on quality or cost, but we are deadly serious about our service proposition. We have to put more thought into all our decisions because, unlike some of our competitors, we have fewer people, pockets that aren’t as deep, and we don’t always have the option to go back to the well. Everything we do has to set us up to move forward quickly, and we must get it right the first time. Partnering with HPE ensures we can deliver.
My bottom line is all about money in, money out. Per our board’s instructions, my objective is to keep the business growing without increasing the headcount. As a company that sells automation, that’s the best affirmation of our services. My headcount is the same as it was a few years ago. We’re cutting costs, but we’ve managed to reduce downtime, outages, and the resulting time spent on customer complaints. We must be doing something right.