nPlan: Minimizing risks for infrastructure schemes around the world with Google Cloud
About nPlan
Founded in 2017, nPlan’s machine learning software draws on one of the world’s largest construction datasets to apply data-driven risk analysis to construction and infrastructure projects. nPlan helps its global clientbase better address risks, save millions of dollars, and significantly reduce delays.
Tell us your challenge. We're here to help.
Contact usTo help its growing network of global infrastructure clients forecast and prepare for risk, nPlan supports its machine learning platform with GKE’s scalable, flexible containerized applications.
Google Cloud results
- Time to identify issues cut by half
- Performance multiplies with 15x more vCPUs in use now
- Lifts barriers to international scaling with a global cloud and support network
- Avoids need for in-house DevOps team as Google Kubernetes Engine (GKE) provides managed services
- Enhances visibility across the operation with a quick and easy to read overview
Scaling an infrastructure risk management tool globally
Whether a tunnel is being dug under a city or a hydroelectric dam is being built across a river, hundreds of unknown variables impact infrastructure projects. Crucial materials could get held up in the supply chain, for example, or an item of archaeological interest could be unearthed. To help organizations manage and plan for a multitude of risks, nPlan uses data-derived forecasting to model the outcomes of future projects.
"Our clients include energy companies, government agencies, and engineering and procurement contractors. Each of their plans could include 100,000 activities, all of which interact with each other and we forecast the risk on every activity in those plans," explains Alan Mosca, Co-Founder and CTO at nPlan. "We then consider where the hotspots are and highlight where extra care must be taken and we provide guidance on how to reduce the uncertainty of producing that project with the intended outcome."
nPlan’s machine learning software draws on one of the world’s largest construction datasets, which is continuously growing as information is added from new projects. And, nPlan’s success is dependent upon having a secure, scalable, flexible solution for storing this information. "We are a data storage and computer power-intensive startup. We don’t have the capacity or budget to maintain our own servers or the space to store them," explains Mosca. "We wouldn’t have gotten off the ground without a versatile cloud storage solution, and we chose Google Cloud because it's global and we’re taking the business global."
"We wouldn’t have gotten off the ground without a versatile cloud storage solution, and we chose Google Cloud because it's global and we’re taking the business global."
—Alan Mosca, Co-Founder and CTO, nPlan24/7 worldwide support was key to nPlan’s international growth plans
Launched in 2017, nPlan is helping a growing list of international clients save time and costs. nPlan customer Network Rail, a British rail operator, for example, estimates that cost savings of up to £30 million could have been achieved on its Great Western Main Line project.
nPlan’s early success has generated a lot of new business and the company opened a Sydney office in 2021 to cope with the global demand for its services. To that end, it was not only looking for a data storage solutions partner with international reach, customer service was just as important.
"We’re still an early-stage startup," explains Mosca, "so we needed a partner that would provide the highest level of service and support and help us to build our systems faster and get our product to market. Google Cloud offers all this, plus an ever-growing offering of tools and apps."
Scaling up supported by Google Kubernetes Engine
Liam Walmsley-Eyre, Engineering Director, joined nPlan in 2019, and to support the company’s growth trajectory he encouraged the team to make better use of the capabilities offered by Google Cloud by moving over to Google Kubernetes Engine (GKE). "The strongest reason for adopting GKE was its ability to scale without friction in response to surges in customer demand," Walmsley-Eyre says. "We had been using automated tools to provision virtual machines, but this was time and labor-consuming. The GKE Horizontal Pod Autoscaler scales things up for you automatically, it does the work for us, something you don’t find in open source components." When a workload is initially moved over to a Kubernetes cluster, it can be difficult to determine resource requirements. By using the Horizontal Pod Autoscaler, nPlan can set the cluster node pools to automatically increase and decrease based on its workload. While nPlan uses another GKE feature, Ingress, to balance loads and route traffic around within clusters.
"We had been using automated tools to provision virtual machines, but this was time and labor-consuming. The GKE Horizontal Pod Autoscaler scales things up for you automatically, it does the work for us, which is something you don’t find in open source components."
—Liam Walmsley-Eyre, Engineering Director, nPLanSeamless migration with Google Kubernetes Engine
As nPlan’s various deployments built up, Walmsley-Eyre also looked to migrate nPlan from a monolithic application structure to a microservices-based one. "We wanted something to lift up the layers of abstraction and manage the underlying operations for us. What’s great about GKE is that it allowed us to do this in silos; we didn’t have to do everything at once, which could have caused disruptions for our customers," he explains.
nPlan was operating on a series of virtual containerized machines and the team was able to lift and shift each container across to GKE. The migration took two months and then attention was shifted to exploring the tools that GKE offers.
"We’re gradually using more and more of the capabilities offered by GKE," says Walmsley-Eyre. "Google Cloud Operations, which collects metrics and logs, is super useful for understanding how our software is performing, why something's going wrong, and what to prioritize for development. I also like Google-managed SSL certificates, because we don’t have to worry about managing certificates now." These domain validation certificates allow Google to provide, renew, and manage domain names.
nPlan needed a single flexible platform to store its ever-increasing body of data gleaned from its machine learning, so it turned to scaleable BigQuery to hold and centralize its live and historical datasets.
Providing faster and easier access to data across the organization
With all its data sitting on Google Cloud, it’s become easier to see what is happening across nPlan’s operations, adds Walmsley-Eyre. "Once fully migrated we had full visibility over all our data. Whether you need to get into the system for maintenance, support or operational purposes, you can quickly see where to look and how to investigate any problems." Consequently, the time it takes nPlan to identify issues has halved and performance has improved by 15X because of the number of vCPUs now in operation.
Meanwhile, centralizing data on a single cloud platform makes testing and development easier and faster for nPlan’s developers too, explains Mosca: "We deploy much faster now, because less time is spent fiddling with how a particular Docker container is running on a VM or making sure our Nginx is routing traffic, for example, which is now all handled by fully-managed services on GKE, which helps minimize overhead efforts."
"We’re now focusing on expansion. GKE will play a big role in this because it makes it easier to manage projects and replicate existing technological infrastructure. It also allows us to dig deeper into the science of risk forecasting and develop new metrics to create even more accurate simulations."
—Alan Mosca, Co-Founder and CTO, nPlannPlan’s ambitious expansion is powered by Google Kubernetes Engine
To meet demand, nPlan’s deployments are scaling fast and have doubled its workforce since the beginning of 2021. And, while you can never completely erase the risks inherent in any business, nPlan’s success proves that they can be reduced. The company forecasts that by integrating Google Cloud into its solutions it will reduce its own liabilities and set itself up for further growth.
Mosca concludes: "We’re now focusing on expansion. GKE will play a big role in this because it makes it easier to manage projects and replicate existing technological infrastructure at scale. It also allows us to dig deeper into the science of risk forecasting and develop new metrics to create even more accurate simulations."
Tell us your challenge. We're here to help.
Contact usAbout nPlan
Founded in 2017, nPlan’s machine learning software draws on one of the world’s largest construction datasets to apply data-driven risk analysis to construction and infrastructure projects. nPlan helps its global clientbase better address risks, save millions of dollars, and significantly reduce delays.