LSE:WNWD is a pioneering Maritime AI company, providing an integrated platform for threat management and maritime domain awareness to accelerate global trade. Global marine activity insights: Windward tracks over 500,000 vessels worldwide, providing real-time data on their daily movements, destinations, cargo transported, weather resilience, and port visitations. While this data is crucial for safeguarding the global supply chain in real-time, it can be challenging to decipher and respond to promptly. With Windward’s comprehensive solution, you’ll gain valuable insights to optimize operations through real-time ETA tracking, enhanced provider efficiency analysis, proactive threat monitoring, and swift mitigation strategies – empowering your organization to navigate the complexities of today’s logistics landscape with confidence.
In 2022, Windward undertook a series of transformations to its utility infrastructure, necessitating a comprehensive review and reassessment of its foundational data architecture. The corporation resolved to invest in enhancing relationships with partners, leveraging data from suppliers, carriers, governments, and insurance companies to integrate maritime intelligence into their internal processes and operations seamlessly? The system empowered all gamers to harness maritime data in unique ways: insurance companies determined value and assessed risk, while governments monitored illicit activities. Therefore, Windward required a foundational data infrastructure built upon an Application Programming Interface (API) driven approach.
As shipping companies navigate increasingly complex sanctions and regulations, Windward has enhanced its AI-driven insights to identify not only potential risks from but also those stemming from, which can obscure the transportation of sanctioned Russian oil or other sensitive commodities. To facilitate rapid innovation and development of additional AI capabilities, Windward’s information platform aimed to enable swift iteration.
To conclude, Windward required a seamless transition of its entire platform from a batch-based data infrastructure to a real-time, streaming architecture. This transition enables users to adapt to novel situations that demand expedited analysis of events previously unforeseen.
On this blog post, we’re excited to introduce Windward’s innovative information platform, built on a pioneering API-first architecture that enables rapid product innovation and empowers real-time, streaming data insights.
Knowledge Challenges
Windward tracks vessel positions derived from Automatic Identification System (AIS) transmissions across the ocean. More than 100 million Automatic Identification System (AIS) transmissions are incrementally recorded daily to accurately track a vessel’s position at any moment in time. When a vessel executes a flip maneuver, Windward can exploit a limited range of AIS transmissions to track its trajectory accurately. This data enables the calculation of velocity, tracks port stops, and identifies various factors that contribute to the overall journey. The AIS transmission data is inconsistent, hindering the assignment of transmissions to the correct vessel. As a result, roughly one-third of all data ultimately necessitates updates and eliminations.
With AIS transmissions providing foundational data, additional enrichment comes from diverse sources, including climate patterns, nautical charts, and other relevant factors such as possession and territorial claims. The enrichment information features evolving schema structures, compounded by the continuous introduction of novel data providers, thereby hindering Windward’s ability to effectively utilize relational databases with rigid schema constraints.
Windward leverages real-time and historical data to conduct comprehensive behavioural evaluations of maritime activities, examining financial performance and detecting potential instances of deceptive transportation practices. Additionally, they develop AI fashion models that enable the detection of environmental threats, sanctions compliance risks, operational hazards, and more. The AI-driven insights initiative prompted Windward to reassess and revamp its entire information architecture, sparking a transformative journey.
As Windward leveraged a batch-based information architecture, they stored raw data in Amazon’s scalable object storage solution, S3. They leveraged MongoDB as their go-to repository for storing and managing vessel and company metadata.
A time-sequenced geospatial data set, comprising vessel position information, was stored in both PostgreSQL and Cassandra databases to facilitate diverse usage scenarios. Windward leveraged specialized databases such as Elasticsearch to enhance performance capabilities, including efficient textual content search functionality. As Windward assessed its information architecture, five disparate databases hindered efforts to facilitate novel usage scenarios, execute efficient contextual queries, and scale database methodologies effectively?
As Windward introduced novel usage scenarios, they encountered constraints with their data infrastructure. Within the phrases of Benny Keinan, Vice President of R&D at Windward, “We have been caught on function improvement and dealing too exhausting on options that ought to have been straightforward to construct. Twelve years ago, our initial foundation of information and mannequin for Windward didn’t meet the advanced search and analytical requirements needed to digitally transform the maritime industry.
Benny and his team were resolute in their decision to develop a cutting-edge information platform designed to better serve the needs of their clients and propel the maritime industry forward. To kick-start their strategy, they started evaluating novel product demands from potential clients and identifying those that would be challenging to support within their current infrastructure, thereby constraining opportunities for generating substantial new revenue streams. These included:
- Geospatial queries: Customers require tailored polygon creation to monitor distinct maritime regions of interest. Their objective was to develop the capability to rapidly search through archives of previously defined polygons and retrieve results in mere seconds, thereby facilitating efficient decision-making and data-driven insights.
- Clients require access to comprehensive vessel information, including contextual data, AIS transmissions, ownership, and action sequences, enabling them to locate specific vessels and track their movements. Users’ search efforts are becoming increasingly arduous to facilitate timely responses within our technological expertise?
- The shopper’s inquiry might require a precise match on a partial vessel identifier, thus the database should efficiently facilitate searches for incomplete phrases.
To enhance their product development prospects, Windward recognized the importance of a database that effectively supports both search and analytics functions, providing real-time access to streaming data.
Necessities for Subsequent-Era Database
To address the complexities arising from disparate databases under management, coupled with the need to support emerging business requirements, Windward chose to unify its data architecture. Taking a user-centric approach, Windward was well-positioned to identify the next essential requirements:
Following a thorough evaluation process, we found that more than 10 databases were assessed, ultimately identifying Rockset and Snowflake as the only two capable of meeting the primary requirements for search and analytics in our application.
As a cloud-based data warehousing platform, Rockset was shortlisted for its unique ability to accelerate analytics on real-time streaming data through its API-first approach and lightning-fast query performance. What’s more, Rockset facilitates seamless in-place updates, ensuring a sustainable approach to processing adjustments to AIS transmissions and the vessels they’re tied to. With expertise in harnessing the power of SQL for complex semi-structured data, Windward recognized an opportunity to integrate geospatial and temporal information into a single, queryable system. One significant limitation of current approaches is their inability to facilitate rapid query processing, prompting Windward’s preference for Rockset, whose innovative indexing mechanism leverages a searchable repository, organized by columnar and row retailers, to support diverse query patterns seamlessly.
Was evaluated for its ability to process and analyze large-scale data sets with ease, leveraging its expertise in columnar storage and advanced skills in executing complex queries, including aggregations and joins, on historical datasets with remarkable speed and efficiency. Each Snowflake and RocksSet are cloud-native and fully managed, thereby reducing the scope of infrastructure operations for the Windward engineering team, allowing them to focus on developing innovative AI-powered insights and capabilities for their maritime platform.
What’s driving the growth in cloud-based data warehousing?
Windward assessed the query efficiency of various methods using a diverse set of six typical queries, in addition to search, geosearch, fuzzy matching, and large-scale aggregation techniques, all executed on a massive 2 billion-record dataset.
The efficacy of Rockset was assessed on a large-scale digital event hosted in the AWS US-West region, utilizing a virtual CPU allocation of 32 units and 256 gigabytes of RAM, which translates to a cost of approximately $7.3496 per hour. The efficiency of Snowflake was assessed against a massive digital repository hosted on Amazon Web Services’ (AWS) US-West region, which incurred a cost of $16 per hour.
Both Rockset and Snowflake claim to offer high-performance, scalable data warehousing capabilities. However, their approaches differ significantly.
The preliminary assessments indicate that Rockset is poised for prompt achievement. While Rockset was considered over Snowflake for Windward’s specific use case, the decision ultimately hinged on various factors that couldn’t be captured by a simplistic comparison. The primary advantage of Rockset’s velocity and performance stems from its Converged Indexing technology, which enables the simultaneous leveraging of numerous indexes in parallel, thereby delivering exceptional speed and efficiency when handling massive datasets?
The thorough efficiency testing conducted by Windward has instilled confidence that Rockset can effectively meet the desired latency requirements, while remaining financially sustainable in both the short and long term.
Navigating the vast expanse of human understanding?
As the maritime ecosystem evolves rapidly, Windward is poised with Rockset to empower its customers by providing real-time visibility and AI-driven intelligence, enabling them to respond nimbly and maintain compliance.
The analytic capabilities, previously requiring significant time and resources to impact Windward’s PostgreSQL database – taking it offline for extended periods or necessitating an estimated 40-minute loading time – are now seamlessly integrated, providing rapid insights to prospective clients in mere seconds. As a result, Windward is streamlining its database management by integrating three existing systems into Rockset, thereby simplifying operations and facilitating the introduction of new products with greater ease. By leveraging this opportunity, Windward’s engineering team gains valuable time to explore and develop innovative AI-driven solutions.
Here is the rewritten text:
“Benny Keinan highlights the transformative impact of Rockset on product improvement, noting that it has enabled his company to offer novel capabilities to customers that were previously unattainable.” As a result, maritime leaders are leveraging AI-driven insights to successfully navigate their supply chains amidst global challenges such as the COVID-19 pandemic, conflicts in Ukraine, and decarbonization efforts. Through Rockset’s innovative technology, we’ve effectively managed the shifting demands of the maritime industry with real-time agility.
You may be enlightened further regarding the fundamental concepts and principles underlying Windward’s AI on their blog site.
