What this blog covers:
- Find out how Kyvos compares with Starburst as an end-to-end analytics platform.
- Read what difference Kyvos AI-powered semantic layer creates in data modeling compared to Starburst that has severe limitations.
- Understand how Kyvos outranks Starburst on different counts, such as querying costs, smart aggregation, speed and scalability.
Starburst, formerly known as Presto SQL, is an end-to-end analytics platform that is being hailed as a full-featured open data lakehouse platform built on open-source Trino. However, the solution has significant limitations when it comes to delivering interactive analysis, performance and semantic layer functionalities and comes at four to five times higher the cost.
An efficient analytics and data management platform should have a semantic layer that simplifies analytics, establishes data trust and improves query performance without constraints of cost, speed or scale. Kyvos is an AI-powered semantic layer that has been designed to meet these goals and has several advantages. The following pointers demonstrate these key differentiators between the two platforms.
Query Performance
Starburst suffers from scalability and low performance issues, especially during high concurrency. This is particularly detrimental when enterprise data volumes and complexities increase. When complex queries involving multiple joints at runtime occur, Starburst’s performance further declines, leading to additional costs and slower responses.
Kyvos can process a query in one second for a single user with 1.5 TB data, while Starburst takes 15 seconds for a single user with 1TB data. Starburst is limited by memory whereas Kyvos delivers interactive query responses, even with more than 1000 concurrent users.
Querying Cost
When it comes to large volumes of data or complex queries, Starburst needs high-configuration nodes. Each of these nodes needs higher memory levels and compute capacity. This in turn needs costlier deployment. For instance, for 1-1.5 TB data size, Starburst will cost four or five times higher than Kyvos.
On the other hand, Kyvos offers price-performant querying models to help scale analytics without additional costs. The platform processes data before feeding it into data models. These optimized models help users save computation expenses at runtime by allowing numerous queries. Enterprises are free to add any number of users to shoot any number of queries without budget escalations. Also, it shuts down the query engines automatically when not needed and restarts them on demand.
A Full-Capacity Semantic Layer
Kyvos is a full-fledged semantic layer that’s capable of delivering complex analytics without the need for technical expertise. It can handle large data models with no limitations on the number of fields, rows and levels of drill-downs. The layer can manage complex and advanced hierarchies, delivering multidimensional analytics and improving query performance.
Through an AI-assisted drag-and-drop visual designer, Kyvos simplifies data modeling and standardizes business logic across the organization. Its semantic layer acts as a single source of truth and unifies data views across any data source, BI tool or cloud platform.
Starburst builds datasets for specific purposes and some of its data products partially function as a semantic layer. As a result, it does not function at par with Kyvos when it comes to handling billion-scale datasets for all departments and users. Enterprises using Starburst need to create multiple SQL queries and views if they wish to support data at scale, which leads to high maintenance costs.
Smart Aggregation Capabilities
Kyvos Smart Recommendations Engine leverages machine learning and advanced algorithms to generate massively scalable data models. These models once created can be used multiple times to process large and complex datasets to deliver sub-second responses. By processing the data in advance, it significantly reduces the response times of all types of queries, be it standard, ad-hoc, old or new.
Starburst doesn’t have the smart aggregation technology; it supports materialized views, that too only for Hive and Iceberg data sources. As a result, enterprises using Starburst need to build a large number of aggregate tables, which further makes managing views difficult.
Gen AI Capabilities and Features
Kyvos Copilot offers a comprehensive suite of Gen AI features, such as a KPI designer assistant with a chat-like interface and generative language capabilities. Starburst’s Gen-AI capabilities are limited to supporting SQL but do not extend to MDX query generation. Starburst’s Gen AI abilities further limit its query types to convert text to only SQL query, unlike Kyvos that supports both SQL and MDX.
Conclusion
Like Starburst, many BI analytics solutions come with a heavy cost without promising seamless data modeling and interpretation. As an in-memory solution, Starburst fails to deliver interactive analysis and falters in dealing with complex data sets and large numbers of concurrent users. Kyvos has been purpose-built to handle the above challenges and simplify analytics while offering speed and scalability at much lower querying costs.
Read the full comparison report to learn more about how Kyvos competes with Starburst.