NAVITIME JAPAN provides a range of navigation services based on its route-search-engine technology. From the total navigation solution NAVITIME to vehicle-specific products like BICYCLE NAVITIME and NAVITIME Drive Supporter, it offers a variety of services for individuals.
In 2016, it entered the travel industry with NAVITIME Travel, which helps users plan their itineraries and book accommodations. In response to increasing inbound demand, it is expanding its presence in the travel field rapidly, including localization into multiple languages. NAVITIME JAPAN's lineup of corporate services is similarly broad, including Business NAVITIME Fleet Management for distribution companies and NAVITIME ASP, which lets customers embed navigation technology in their websites to guide visitors to stores.
With more than 40 million users every month and about 4.8 million paying users as of December 2017, NAVITIME JAPAN decided to switch from its on-premises infrastructure to a cloud solution in light of its expanding user base. "Our route search services see very different usage patterns depending on the season and the weather. We had simply reached the limit of our ability to procure and operate on-premises servers to meet the anticipated peak demand. We chose AWS for its flexibility in the face of changing access patterns," says NAVITIME Executive Vice President and CTO, Shin Kikuchi.
NAVITIME JAPAN's log analysis system, which drew on location information, trajectory data, and route search information, also ran on its on-premises environment. The company found it difficult to maintain the storage capacity and processing ability needed to support its up to 300 GB of logs each day. To increase its flexibility, NAVITIME JAPAN switched to Amazon Simple Storage Service (Amazon S3) for storage and managed Hadoop service Amazon EMR for analysis. Initially, they then migrated their analysis engine to another cloud provider's big data query service, transmitting the necessary log data from Amazon S3.
This approach made log aggregation faster, but also created new issues, including the cost of sending data and storing two copies, security and access concerns, and increased operational workload. "Sending the log data from S3 to the query service for analysis meant we were paying transmission costs and additional storage costs. "But security was an even bigger problem. The log data was sent over the internet, and the query service didn't allow fine-grained permissions management. We needed a better solution," says Kazuki Tanaka, cloud lead and infrastructure engineer at NAVITIME.
NAVITIME JAPAN was still searching for the solution to these problems when AWS announced its serverless query service, Amazon Athena, at AWS re:Invent 2016. The decision to adopt Athena was immediate, says Infrastructure Engineer Kazuhiro Shintate. "The deciding factor was the opportunity to resolve our log-data transmission cost problem and security issues all at once. Because it was a managed service, we also anticipated lower operational workload. Administrating would be simpler if we could use Amazon S3 as a single data lake. Finally, we appreciated that it was an AWS service since we were familiar with their products and had successfully used them in the past."
The current system takes log data in bulk, anonymizes it using EMR, and then saves it to S3 for analysis by Athena. "Because S3 and Athena use the same infrastructure, there's no need to transmit data or store it at the other end. This reduced our infrastructure costs by 75 percent overall. We used to spend eight hours transmitting data every day, which meant that log information wasn't available until two days later. Now, that information is available the following day. We have also implemented more fine-grained access permissions," says Tanaka.
Adopting Athena had another benefit: it allowed NAVITIME JAPAN to release Road Profiler, a corporate traffic analysis service. Road Profiler uses mobile GPS probe data obtained from products like CAR NAVITIME and NAVITIME Drive Supporter.
CAR NAVITIME combines this data with attributes from app users to analyze routes and congestion on roads nationwide and aggregate and visualize this information. "We tried many options for this service, but only Athena was able to meet the analysis response requirements. Without the serverless, scalable service provided by Athena, we wouldn't have been able to offer Road Profiler at all," says Shintate.
Deployment brought other benefits as well. The company's infrastructure department no longer has difficulty attracting talent. "When our system was on premises, the infrastructure department's image was dominated by the idea of servers that needed to be watched 24 hours a day. Hardly anyone asked to be assigned there. The idea of moving to the cloud on AWS was an immediate motivation boost. It drew so much interest from engineers in the infrastructure department and so many requests for reassignment there that it resolved the department's personnel difficulties entirely," says Kikuchi.
A new cost-consciousness among engineers is another benefit the company has realized. "In the past, employees simply procured and paid for servers according to the department's needs. With AWS, monthly costs are visible, which makes cost-consciousness possible. As a result, the route-search-engine development division was able to cut costs by around 20 percent by improving server performance," says Kikuchi.
NAVITIME is also taking full advantage of AWS Enterprise Support to receive needed support, including architecture reviews from AWS Solutions Architects. "From clearing up minor uncertainties to discussing changes in specifications, we have received responsive and helpful answers to a range of questions. They have also offered advice on architecture based on the AWS Well-Architected Framework and information about new services. We have found them to be extremely helpful," says Tanaka.
Going forward, NAVITIME plans to refine its core route-search-engine technology, offering faster, more accurate search services through Amazon Elastic Compute Cloud (Amazon EC2) GPU technology. "We're approaching the limit of what can be achieved by a CPU-based system. To offer superior services in future, we're developing GPU-based engine algorithms. We plan to use Amazon EC2 P3 instances and other services offering GPU technology from for this. By evolving our architecture alongside these efforts, we expect to deploy a service that our customers can use within fiscal 2018," says Kikuchi.