Jonathan Lalou's blog

[Voxxed Amsterdam 2025] From Zero to AI: Building Smart Java or Kotlin Applications with Spring AI

At VoxxedDaysAmsterdam2025, Christian Tzolov, a Spring AI team member at VMware and lead of the MCP Java SDK, delivered a comprehensive session titled “From Zero to AI: Building Smart Java or Kotlin Applications with Spring AI.” Spanning nearly two hours, the session provided a deep dive into integrating generative AI into Java and Kotlin applications using Spring AI, a framework designed to connect enterprise data and APIs with AI models. Through live coding demos, Tzolov showcased practical use cases, including conversation memory, tool/function calling, retrieval-augmented generation (RAG), and multi-agent systems, while addressing challenges like AI hallucinations and observability. Attendees left with actionable insights to start building AI-driven applications, leveraging Spring AI’s portable abstractions and the Model Context Protocol (MCP).

Overcoming LLM Limitations with Spring AI

Tzolov began by outlining the challenges of large language models (LLMs): they are stateless, frozen in time, and lack domain-specific knowledge, requiring developers to provide context, manage state, and handle interactions with external systems. Spring AI addresses these issues with high-level abstractions like the ChatClient, similar to Spring’s RestClient or WebClient, enabling seamless integration with models like OpenAI’s GPT-4o, Anthropic’s Claude, or open-source alternatives like LLaMA. A live demo of a flight booking assistant illustrated these concepts. Tzolov started with a basic Spring Boot application connected to OpenAI, demonstrating a simple chat interface. To ground the model, he used system prompts to define its behavior as a customer support agent for “Fun Air,” ensuring contextually appropriate responses. He then introduced conversation memory using Spring AI’s ChatMemoryAdvisor, which retains a chronological list of messages to maintain state, addressing the stateless nature of LLMs. For long-term memory, Tzolov employed a vector store (Chroma) to store conversation history semantically, retrieving only relevant data for queries, thus overcoming context window limitations. This setup allowed the assistant to respond accurately to queries like “What is my flight status?” by fetching booking details (e.g., booking number 103) from a mock database.

Enhancing AI Applications with Tool Calling and RAG

To enable LLMs to interact with external systems, Tzolov demonstrated tool/function calling, where Spring AI wraps existing services (e.g., a flight booking service) as tools with metadata (name, description, JSON schema). In the demo, the assistant used a getBookingDetails tool to query a database, allowing it to provide accurate flight status updates. Tzolov emphasized the importance of descriptive tool metadata to guide the LLM in deciding when and how to invoke tools, reducing the risk of misinterpretation. For domain-specific knowledge, he introduced prompt stuffing—injecting additional context into prompts—and RAG for dynamic data retrieval. In a RAG demo, cancellation policies were loaded into a Chroma vector store, chunked into meaningful segments, and retrieved dynamically based on user queries. This approach mitigated hallucinations, as seen when the assistant correctly cited a 50% refund policy for premium economy bookings within 40 hours. Tzolov highlighted advanced RAG techniques, such as data compression and reranking, supported by Spring AI’s APIs, and stressed the importance of evaluating responses to ensure relevance, referencing frameworks like those from contributor Thomas Vitali.

Building Multi-Agent Systems with MCP

Tzolov explored the Model Context Protocol (MCP), initiated by Anthropic, as a standardized way to integrate AI applications with external tools and resources across platforms. Using Spring AI’s MCP Java SDK, he demonstrated how to build and consume MCP-compliant tools. In one demo, a Spring AI application connected to MCP servers for Brave Search (JavaScript-based) and file system access, enabling an agent to answer queries about Spring AI support for MCP and write summaries to a file. Another demo reversed the setup, exposing a Spring AI weather tool (using Open-Meteo) as an MCP server, accessible by third-party clients like Claude Desktop via standard I/O or HTTP/SSE transports. Tzolov explained MCP’s bidirectional architecture, where clients can act as servers, supporting features like sampling (allowing servers to request LLM processing from clients). He addressed security concerns, noting Spring AI’s integration with Spring Security (referencing a blog by Daniel Garnier-Moiroux) to secure MCP servers with OAuth 2.1. The session also introduced agentic systems, where LLMs act as a “brain” for planning and tools as a “body” for interaction, with an agentic loop evaluating and refining responses. A work-in-progress demo showcased an orchestration pattern, delegating tasks to searcher, fact-checker, and writer agents, to be published on the Spring AI Community Portal.

Observability and Multimodality for Robust AI Systems

Observability was a key focus, as Tzolov underscored its importance in debugging complex AI interactions. Spring AI integrates with Micrometer to provide metrics (e.g., token usage, model throughput, latency), tracing, and logging (via Loki). A dashboard demo displayed real-time metrics for the flight booking assistant, highlighting tool calls and errors, crucial for diagnosing issues in agentic systems. Tzolov also explored multimodality, demonstrating a voice assistant using OpenAI’s GPT-4o audio preview, which processes audio input and output. Configured as “Marvin the Paranoid Android,” the assistant responded to voice queries with humorous, contextually appropriate replies, showcasing Spring AI’s support for non-text modalities like images, PDFs, and videos (e.g., Gemini’s video support). Tzolov noted that multimodality enables richer interactions, such as analyzing images or converting PDFs to markdown, and Spring AI’s abstractions handle these seamlessly. He concluded by encouraging developers to explore Spring AI’s documentation, experiment with MCP, and contribute to the community, emphasizing its role in building robust, interoperable AI applications.

Links:

Hashtags: #SpringAI #GenerativeAI #ModelContextProtocol #ChristianTzolov #VoxxedDaysAmsterdam2025 #AIAgents #RAG #Observability

Posted in en-US | Tags: AI, LLM, Spring AI | No Comments »

[Voxxed Amsterdam 2025] How to Survive as a Developer in the Exponential Age of AI

Author: Jonathan Lalou

In a dynamic and fast-paced session at VoxxedDaysAmsterdam2025, Sander Hoogendoorn, CTO at iBOOD.com, explores the transformative impact of artificial intelligence (AI) on software development. With over four decades of coding experience, Hoogendoorn demystifies the hype around AI, examining its practical benefits, challenges, and implications for developers. Far from signaling the end of programming careers, he argues that AI empowers developers to tackle broader and more complex problems—provided they adapt and remain committed to quality and lifelong learning.

AI: A Developer’s Ally

Hoogendoorn clarifies AI’s role in development, highlighting tools like Cursor AI, which can extract components, fix linting issues, and generate unit tests using natural language prompts. He recounts an experience where his non-technical colleague Elmo built an iOS app with Cursor AI, illustrating AI’s democratizing potential. However, Sander warns that such tools require supervision to ensure code quality and compliance with project standards. AI’s ability to automate repetitive tasks—such as refactoring or test generation—frees developers to focus on complex problem-solving. At iBOOD, AI has enabled the team to create content with a unique tone, retrieve competitor pricing, and automate invoice recognition—tasks that previously required external expertise or significant manual effort.

Navigating Technical Debt in the AI Era

The rise of AI-assisted development—especially “vibe coding,” where problems are described in natural language to generate code—introduces new forms of technical debt. Hoogendoorn references Ward Cunningham’s metaphor of technical debt as a loan that accelerates development but demands repayment through refactoring. AI-generated code, while fast to produce, often lacks context or long-term maintainability. For instance, Cursor AI struggled to integrate with iBOOD’s custom markdown components, resulting in complex solutions that required manual fixes. Research suggests that AI can amplify technical debt if used without rigorous validation, emphasizing the need for developers to stay vigilant and prioritize code quality over short-term gains.

Thriving in an AI-Centric Future

Far from replacing developers, Sander Hoogendoorn asserts that AI enhances their capabilities, enabling them to tackle ambitious challenges. He reminds us that developers are not mere typists—they are problem-solvers who think critically and collaborate to meet business needs. Historical shifts, from COBOL to cloud computing, have always empowered developers to solve bigger problems, and AI is no exception. By thoughtfully experimenting with AI—integrating it into workflows for content creation, price retrieval, or invoice processing—Sander’s team at iBOOD has unlocked previously unreachable efficiencies. The key to thriving, he says, lies in relentless learning and a willingness to adapt, ensuring that developers remain indispensable in an AI-driven world.

Links:

Hashtags: #AI #SoftwareDevelopment #TechnicalDebt #CursorAI #iBOOD #SanderHoogendoorn #VoxxedDaysAmsterdam2025

Posted in en-US | Tags: AI, Voxxed | No Comments »

[Oracle Dev Days 2025] Optimizing Java Performance: Choosing the Right Garbage Collector

Author: Jonathan Lalou

Jean-Philippe BEMPEL , a seasoned developer at Datadog and a Java Champion, delivered an insightful presentation on selecting and tuning Garbage Collectors (GCs) in OpenJDK to enhance Java application performance. His talk, rooted in practical expertise, unraveled the complexities of GCs, offering a roadmap for developers to align their choices with specific application needs. By dissecting the characteristics of various GCs and their suitability for different workloads, Jean-Philippe provided actionable strategies to optimize memory management, reduce production issues, and boost efficiency.

Understanding Garbage Collectors in OpenJDK

Garbage Collectors are pivotal in Java’s memory management, silently handling memory allocation and reclamation. However, as Jean-Philippe emphasized, a misconfigured GC can lead to significant performance bottlenecks in production environments. OpenJDK offers a suite of GCs—Serial GC, Parallel GC, G1, Shenandoah, and ZGC—each designed with distinct characteristics to cater to diverse application requirements. The challenge lies in selecting the one that best matches the workload, whether it prioritizes throughput or low latency.

Jean-Philippe began by outlining the foundational concepts of GCs, particularly the generational model. Most GCs in OpenJDK are generational, dividing memory into the Young Generation (for short-lived objects) and the Old Generation (for longer-lived objects). The Young Generation is further segmented into the Eden space, where new objects are allocated, and Survivor spaces, which hold objects that survive initial collections before promotion to the Old Generation. Additionally, the Metaspace stores class metadata, a critical but often overlooked component of memory management.

Serial GC: Simplicity for Constrained Environments

The Serial GC, one of the oldest collectors, operates with a single thread and employs a stop-the-world approach, pausing all application threads during collection. Jean-Philippe highlighted its suitability for small-scale applications, particularly those running in containers with less than 2 GB of RAM, where it serves as the default GC. Its simplicity makes it ideal for environments with limited resources, but its stop-the-world nature can introduce noticeable pauses, making it less suitable for latency-sensitive applications.

To illustrate, Jean-Philippe explained the mechanics of the Young Generation’s Survivor spaces. These spaces, S0 and S1, alternate roles as source and destination during minor GC cycles, copying live objects to manage memory efficiently. Objects surviving multiple cycles are promoted to the Old Generation, reducing the overhead of frequent collections. This generational approach leverages the hypothesis that most objects die young, minimizing the cost of memory reclamation.

Parallel GC: Maximizing Throughput

For applications prioritizing throughput, such as batch processing jobs, the Parallel GC offers significant advantages. Unlike the Serial GC, it leverages multiple threads to reclaim memory, making it efficient for systems with ample CPU cores. Jean-Philippe noted that it was the default GC until JDK 8 and remains a strong choice for throughput-oriented workloads like Spark jobs, Kafka consumers, or ETL processes.

The Parallel GC, also stop-the-world, excels in scenarios where total execution time matters more than individual pause durations. Jean-Philippe shared a benchmark using a JFR (Java Flight Recorder) file parsing application, where Parallel GC outperformed others, achieving a throughput of 97% (time spent in application versus GC). By tuning the Young Generation size to reduce frequent minor GCs, developers can further minimize object copying, enhancing overall performance.

G1 GC: Balancing Throughput and Latency

The G1 (Garbage-First) GC, default since JDK 9 for heaps larger than 2 GB, strikes a balance between throughput and latency. Jean-Philippe described its region-based memory management, dividing the heap into smaller regions (Eden, Survivor, Old, and Humongous for large objects). This structure allows G1 to focus on regions with the most garbage, optimizing memory reclamation with minimal copying.

In his benchmark, G1 showed a throughput of 85%, with average pause times of 76 milliseconds, aligning with its target of 200 milliseconds. However, Jean-Philippe pointed out challenges with Humongous objects, which can increase GC frequency if not managed properly. By adjusting region sizes (up to 32 MB), developers can mitigate these issues, improving throughput for applications like batch jobs while maintaining reasonable pause times.

Shenandoah and ZGC: Prioritizing Low Latency

For latency-sensitive applications, such as HTTP servers or microservices, Shenandoah and ZGC are the go-to choices. These concurrent GCs minimize pause times, often below a millisecond, by performing most operations alongside the running application. Jean-Philippe highlighted Shenandoah’s non-generational approach (though a generational version is in development) and ZGC’s generational support since JDK 21, making the latter particularly efficient for large heaps.

In a latency-focused benchmark using a Spring PetClinic application, Jean-Philippe demonstrated that Shenandoah and ZGC maintained request latencies below 200 milliseconds, significantly outperforming Parallel GC’s 450 milliseconds at the 99th percentile. ZGC’s use of colored pointers and load/store barriers ensures rapid memory reclamation, allowing regions to be freed early in the GC cycle, a key advantage over Shenandoah.

Tuning Strategies for Optimal Performance

Tuning GCs is as critical as selecting the right one. For Parallel GC, Jean-Philippe recommended sizing the Young Generation to reduce the frequency of minor GCs, ideally exceeding 50% of the heap to minimize object copying. For G1, adjusting region sizes can address Humongous object issues, while setting a maximum pause time target (e.g., 50 milliseconds) can shift its behavior toward latency sensitivity, though it may not compete with Shenandoah or ZGC in extreme cases.

For concurrent GCs like Shenandoah and ZGC, ensuring sufficient heap size and CPU cores prevents allocation stalls, where threads wait for memory to be freed. Jean-Philippe emphasized that Shenandoah requires careful heap sizing to avoid full GCs, while ZGC’s rapid region reclamation reduces such risks, making it more forgiving for high-allocation-rate applications.

Selecting the Right GC for Your Workload

Jean-Philippe concluded by categorizing workloads into two types: throughput-oriented (SPOT) and latency-sensitive. For SPOT workloads, such as batch jobs or ETL processes, Parallel GC or G1 are optimal, with Parallel GC offering easier tuning for predictable performance. For latency-sensitive applications, like microservices or databases (e.g., Cassandra), ZGC’s generational efficiency and Shenandoah’s low-pause capabilities shine, with ZGC being particularly effective for large heaps.

By analyzing workload characteristics and leveraging tools like GC Easy for log analysis, developers can make informed GC choices. Jean-Philippe’s benchmarks underscored the importance of tailoring GC configurations to specific use cases, ensuring both performance and stability in production environments.

Links:

Hashtags: #Java #GarbageCollector #OpenJDK #Performance #Tuning #Datadog #JeanPhilippeBempel #OracleDevDays2025

Posted in en-US | Tags: Garbage Collector, JVM, OpenJDK, Performance, Turning | No Comments »

(temporary testing) Mapping pages

Author: Jonathan Lalou

Posted in en-US | No Comments »

[Oracle Dev Days 2025] From JDK 21 to JDK 25: Jean-Michel Doudoux on Java’s Evolution

Author: Jonathan Lalou

Jean-Michel Doudoux, a renowned Java Champion and Sciam consultant, delivered a session, charting Java’s evolution from JDK 21 to JDK 25. As the next Long-Term Support (LTS) release, JDK 25 introduces transformative features that redefine Java development. Jean-Michel’s talk provided a comprehensive guide to new syntax, APIs, JVM enhancements, and security measures, equipping developers to navigate Java’s future with confidence.

Enhancing Syntax and APIs

Jean-Michel began by exploring syntactic improvements that streamline Java code. JEP 456 in JDK 22 introduces unnamed variables using _, improving clarity for unused variables. JDK 23’s JEP 467 adds Markdown support for Javadoc, easing documentation. In JDK 25, JEP 511 simplifies module imports, while JEP 512’s implicit classes and simplified main methods make Java more beginner-friendly. JEP 513 enhances constructor flexibility, enabling pre-constructor logic. These changes collectively minimize boilerplate, boosting developer efficiency.

Expanding Capabilities with New APIs

The session highlighted APIs that broaden Java’s scope. The Foreign Function & Memory API (JEP 454) enables safer native code integration, replacing sun.misc.Unsafe. Stream Gatherers (JEP 485) enhance data processing, while the Class-File API (JEP 484) simplifies bytecode manipulation. Scope Values (JEP 506) improve concurrency with lightweight alternatives to thread-local variables. Jean-Michel’s practical examples demonstrated how these APIs empower developers to craft modern, robust applications.

Strengthening JVM and Security

Jean-Michel emphasized JVM and security advancements. JEP 472 in JDK 25 restricts native code access via --enable-native-access, enhancing system integrity. The deprecation of sun.misc.Unsafe aligns with safer alternatives. The removal of 32-bit support, the Security Manager, and certain JMX features reflects Java’s modern focus. Performance boosts in HotSpot JVM, Garbage Collectors (G1, ZGC), and startup times via Project Leyden (JEP 483) ensure Java’s competitiveness.

Boosting Productivity with Tools

Jean-Michel covered enhancements to Java’s tooling ecosystem, including upgraded Javadoc, JCMD, and JAR utilities, which streamline workflows. New Java Flight Recorder (JFR) events improve diagnostics. He urged developers to test JDK 25’s early access builds to prepare for the LTS release, highlighting how these tools enhance efficiency and scalability in application development.

Navigating JDK 25’s LTS Future

Jean-Michel wrapped up by emphasizing JDK 25’s role as an LTS release with extended support. He encouraged proactive engagement with early access programs to adapt to new features and deprecations. His session offered a clear, actionable roadmap, empowering developers to leverage JDK 25’s innovations confidently. Jean-Michel’s expertise illuminated Java’s trajectory, inspiring attendees to embrace its evolving landscape.

Links:

Hashtags: #Java #JDK25 #LTS #JVM #Security #Sciam #JeanMichelDoudoux

Posted in en-US | Tags: Java, JDK, JEP, JSR | No Comments »

S	M	T	W	T	F	S
« Jun
		1	2	3	4	5
6	7	8	9	10	11	12
13	14	15	16	17	18	19
20	21	22	23	24	25	26
27	28	29	30	31