Sertis Corp. enables customers to analyze own data using platform running in Google Cloud
Founded in 2014 and headquartered in Bangkok, Thailand, Sertis Corp. provides data science and big data analytics services to businesses in a range of industries. The consultancy runs a platform called ‘Datana’ that incorporates a simple web interface that makes it easy for clients to analyze data. The platform incorporates a range of tools - including big data pipeline integration and provisioning - that enable clients to undertake data ingestion, preparation, modeling and deployment.
“Many businesses do not yet understand the full potential of big data and analytics,” says Nikun Khoongumjorn, VP of Software Engineering, Sertis Corp. “We’re here to help our clients make sense of their data and use it to inform their processes and outcomes.”
Sertis Corp. delivers Datana as a service or in customers’ own data centers. The consultancy has about 50 team members comprising data scientists, data analysts and engineers. The developers build and maintain the platform; engineers typically clean the customers’ data; the data scientists develop models to run the data through; and analysts interpret and record the results and present the analysis to the customers.
The development team’s tasks include ensuring Datana meets the needs of the consultancy’s data scientists. “I’ve come in to lead the development team and help them adopt a more structured approach to designing and building the platform,” says Khoongumjorn. “We’re obtaining input from the data scientists and engineers to ensure we provide exactly what they need. When we’ve fully met their requirements, we’ll start enabling customers to integrate directly to our platform and serve themselves with functionality that meets their needs.”
Sertis Corp. had initially deployed Datana in a public cloud service. However, the cost of the service proved too much as business growth prompted increased use of cloud resources. The consultancy employed workarounds to control its costs, such as undertaking heavy processing activities during the day rather than at night. However, with the cloud resources already in use, this approach created bottlenecks that compromised performance.
The development team started experimenting with Google Cloud Platform (GCP) and the comparatively low costs soon prompted the consultancy to move Datana to the cloud platform. “We needed high-performance, secure machines and networks that could move large volumes of data quickly between them,” says Khoongumjorn. “GCP could meet these requirements.”
Sertis Corp. migrated its platform to GCP in October 2016 and also runs its corporate website and a range of external websites on the platform. The consultancy uses Google Compute Engine virtual machine instances to process datasets, while Google Cloud Storage stores data centrally where Sertis Co. can grant customers access to perform analysis and associated activities. Firebase provides the landing page for the Sertis Corp. platform. The consultancy’s steady-state usage comprises 1,200 virtual CPUs, 4,754 GB RAM and 111 TB of disk capacity, with the number of cores increased in bursts as required to support demand. Sertis Corp. is also using short-lived preemptible virtual machines when appropriate to help control its costs.
Sertis Corp. completed the deployment using internal resources thanks to the ease of use of GCP and the clarity of the cloud platform’s documentation. About 90 percent of the processing, analysis and associated activities the consultancy undertakes for customers runs on GCP, with the remaining processes undertaken on local machines.
Moving to GCP has enabled Sertis Corp. to reduce its costs and operate more effectively. “We were able to reduce our costs by 50 percent by moving to GCP,” says Khoongumjorn. “In addition, GCP performed better against various benchmarks and all its resources were available as we needed them.”
Sertis Co. has accelerated by up to 65 percent the speed that large volumes of data are transferred between instances with GCP, and provides reports to customers about 20 percent faster than on the previous cloud service. Being able to run on GCP 24 hours a day has also sped up tasks during the day and freed up about 10 percent of the time of people charged with monitoring these activities.
Customer service levels have also improved considerably. “Because we can run faster and deeper analytics using GCP, we can now obtain more insightful results for our customers,” says Khoongumjorn. “GCP has provided considerable value to our business – we’re very happy with its capabilities and how we are using the platform now,” he adds.