Support virtual gen ai analysis for otlp and zipkin traces

.github/workflows/skywalking.yaml

@@ -635,6 +635,10 @@ jobs:
             config: test/e2e-v2/cases/zipkin/banyandb/e2e.yaml
           - name: Virtual GenAI
             config: test/e2e-v2/cases/virtual-genai/e2e.yaml
+          - name: OTLP Virtual GenAI
+            config: test/e2e-v2/cases/otlp-virtual-genai/e2e.yaml
+          - name: Zipkin Virtual GenAI
+            config: test/e2e-v2/cases/zipkin-virtual-genai/e2e.yaml
 
           - name: Nginx
             config: test/e2e-v2/cases/nginx/e2e.yaml

docs/en/changes/changes.md

@@ -169,7 +169,7 @@
 * Support Virtual-GenAI monitoring.
 * Fix on-demand pod log parsing failure by replacing invalid `DateTimeFormatter` pattern with `ISO_OFFSET_DATE_TIME`.
 * Fix Zipkin receiver compatibility with application/x-protobuf Content-Type.
-
+* Support virtual GenAI analysis for otlp and zipkin traces.
 
 #### UI
 * Fix the missing icon in new native trace view.

docs/en/setup/service-agent/virtual-genai.md

@@ -6,6 +6,13 @@ metrics of the GenAI operations are from the GenAI client-side perspective.
 For example, a Spring AI plugin in the Java agent could detect the latency of a chat completion request.
 As a result, SkyWalking would show traffic, latency, success rate, token usage (input/output), and estimated cost in the GenAI dashboard.
 
+# Data Sources
+Virtual GenAI metrics are derived from distributed tracing data. SkyWalking OAP can ingest and analyze trace data adhering to GenAI semantic conventions from the following sources:
+
+1. Native SkyWalking Traces via SkyWalking Java Agent
+2. OpenTelemetry format trace 
+3. Zipkin format Traces
+
 ## Span Contract
 
 The GenAI operation span should have the following properties:
@@ -60,4 +67,19 @@ The following metrics are available at the **model** (service instance) level:
 - `gen_ai_model_total_estimated_cost / avg_estimated_cost` - Estimated cost
 
 ## Requirement
-`SkyWalking Java Agent` version >= 9.7
+
+### Version
+`SkyWalking Java Agent` version >= 9.7
+
+### Semantic Conventions and Compatibility
+The tag keys used in Virtual GenAI follow the **OpenTelemetry GenAI Semantic Conventions**. SkyWalking OAP identifies GenAI-related spans based on the following criteria depending on the data source:
+
+* **SkyWalking Native Agent**: Requires an **Exit** span with `SpanLayer == GENAI` and relevant `gen_ai.*` tags.
+* **OTLP / Zipkin Traces**: Any span containing the `gen_ai.response.model` tag will be identified as a GenAI operation.
+
+**Note on OTLP / Zipkin Provider Identification**:
+To ensure broad compatibility with different OpenTelemetry instrumentation versions, SkyWalking OAP identifies the GenAI provider using the following prioritized logic:
+
+1. **`gen_ai.provider.name`**: SkyWalking first looks for this tag (the latest OTel semantic convention).
+2. **`gen_ai.system`**: If the above is missing, it falls back to this legacy tag for backward compatibility with older instrumentation (e.g., current OTel Python auto-instrumentation).
+3. **Prefix Matching**: If neither tag is present, SkyWalking attempts to identify the provider by matching the model name against the `prefix-match` rules defined in the `gen-ai-config.yml`.

oap-server/analyzer/agent-analyzer/src/main/java/org/apache/skywalking/oap/server/analyzer/provider/trace/parser/listener/VirtualServiceAnalysisListener.java

@@ -95,7 +95,7 @@ public AnalysisListener create(ModuleManager moduleManager, AnalyzerModuleConfig
                             new VirtualCacheProcessor(namingControl, config),
                             new VirtualDatabaseProcessor(namingControl, config),
                             new VirtualMQProcessor(namingControl),
-                            new VirtualGenAIProcessor(namingControl, genAIMeterAnalyzerService)
+                            new VirtualGenAIProcessor(genAIMeterAnalyzerService)
                     )
             );
         }

oap-server/analyzer/agent-analyzer/src/main/java/org/apache/skywalking/oap/server/analyzer/provider/trace/parser/listener/vservice/VirtualGenAIProcessor.java

@@ -22,13 +22,7 @@
 import org.apache.skywalking.apm.network.language.agent.v3.SpanLayer;
 import org.apache.skywalking.apm.network.language.agent.v3.SpanObject;
 import org.apache.skywalking.oap.analyzer.genai.service.IGenAIMeterAnalyzerService;
-import org.apache.skywalking.oap.server.core.analysis.Layer;
-import org.apache.skywalking.oap.server.core.config.NamingControl;
 import org.apache.skywalking.oap.server.core.source.GenAIMetrics;
-import org.apache.skywalking.oap.server.core.source.GenAIModelAccess;
-import org.apache.skywalking.oap.server.core.source.GenAIProviderAccess;
-import org.apache.skywalking.oap.server.core.source.ServiceInstance;
-import org.apache.skywalking.oap.server.core.source.ServiceMeta;
 import org.apache.skywalking.oap.server.core.source.Source;
 
 import java.util.ArrayList;
@@ -38,8 +32,6 @@
 @RequiredArgsConstructor
 public class VirtualGenAIProcessor implements VirtualServiceProcessor {
 
-    private final NamingControl namingControl;
-
     private final IGenAIMeterAnalyzerService meterAnalyzerService;
 
     private final List<Source> recordList = new ArrayList<>();
@@ -55,53 +47,7 @@ public void prepareVSIfNecessary(SpanObject span, SegmentObject segmentObject) {
             return;
         }
 
-        recordList.add(toServiceMeta(metrics));
-        recordList.add(toInstance(metrics));
-        recordList.add(toProviderAccess(metrics));
-        recordList.add(toModelAccess(metrics));
-    }
-
-    private ServiceMeta toServiceMeta(GenAIMetrics metrics) {
-        ServiceMeta service = new ServiceMeta();
-        service.setName(namingControl.formatServiceName(metrics.getProviderName()));
-        service.setLayer(Layer.VIRTUAL_GENAI);
-        service.setTimeBucket(metrics.getTimeBucket());
-        return service;
-    }
-
-    private Source toInstance(GenAIMetrics metrics) {
-        ServiceInstance instance = new ServiceInstance();
-        instance.setTimeBucket(metrics.getTimeBucket());
-        instance.setName(namingControl.formatInstanceName(metrics.getModelName()));
-        instance.setServiceLayer(Layer.VIRTUAL_GENAI);
-        instance.setServiceName(metrics.getProviderName());
-        return instance;
-    }
-
-    private GenAIProviderAccess toProviderAccess(GenAIMetrics metrics) {
-        GenAIProviderAccess source = new GenAIProviderAccess();
-        source.setName(namingControl.formatServiceName(metrics.getProviderName()));
-        source.setInputTokens(metrics.getInputTokens());
-        source.setOutputTokens(metrics.getOutputTokens());
-        source.setTotalEstimatedCost(metrics.getTotalEstimatedCost());
-        source.setLatency(metrics.getLatency());
-        source.setStatus(metrics.isStatus());
-        source.setTimeBucket(metrics.getTimeBucket());
-        return source;
-    }
-
-    private GenAIModelAccess toModelAccess(GenAIMetrics metrics) {
-        GenAIModelAccess source = new GenAIModelAccess();
-        source.setServiceName(namingControl.formatServiceName(metrics.getProviderName()));
-        source.setModelName(namingControl.formatInstanceName(metrics.getModelName()));
-        source.setInputTokens(metrics.getInputTokens());
-        source.setOutputTokens(metrics.getOutputTokens());
-        source.setTotalEstimatedCost(metrics.getTotalEstimatedCost());
-        source.setTimeToFirstToken(metrics.getTimeToFirstToken());
-        source.setLatency(metrics.getLatency());
-        source.setStatus(metrics.isStatus());
-        source.setTimeBucket(metrics.getTimeBucket());
-        return source;
+        recordList.addAll(meterAnalyzerService.transferToSources(metrics));
     }
 
     @Override

oap-server/analyzer/gen-ai-analyzer/src/main/java/org/apache/skywalking/oap/analyzer/genai/GenAIAnalyzerModuleProvider.java

@@ -26,6 +26,7 @@
 import org.apache.skywalking.oap.analyzer.genai.service.GenAIMeterAnalyzer;
 import org.apache.skywalking.oap.analyzer.genai.service.IGenAIMeterAnalyzerService;
 import org.apache.skywalking.oap.server.core.CoreModule;
+import org.apache.skywalking.oap.server.core.config.NamingControl;
 import org.apache.skywalking.oap.server.core.oal.rt.OALEngineLoaderService;
 import org.apache.skywalking.oap.server.library.module.ModuleConfig;
 import org.apache.skywalking.oap.server.library.module.ModuleDefine;
@@ -37,6 +38,8 @@ public class GenAIAnalyzerModuleProvider extends ModuleProvider {
 
     private GenAIConfig config;
 
+    private GenAIMeterAnalyzer analyzer;
+
     @Override
     public String name() {
         return "default";
@@ -67,10 +70,11 @@ public void prepare() throws ServiceNotProvidedException, ModuleStartException {
         GenAIConfigLoader loader = new GenAIConfigLoader(config);
         config = loader.loadConfig();
         GenAIProviderPrefixMatcher matcher = GenAIProviderPrefixMatcher.build();
+
+        this.analyzer = new GenAIMeterAnalyzer(matcher);
         this.registerServiceImplementation(
                 IGenAIMeterAnalyzerService.class,
-                new GenAIMeterAnalyzer(matcher)
-        );
+                analyzer);
     }
 
     @Override
@@ -79,6 +83,12 @@ public void start() throws ServiceNotProvidedException, ModuleStartException {
                 .provider()
                 .getService(OALEngineLoaderService.class)
                 .load(GenAIOALDefine.INSTANCE);
+
+        NamingControl namingControl = getManager().find(CoreModule.NAME)
+                .provider()
+                .getService(NamingControl.class);
+
+        this.analyzer.setNamingControl(namingControl);
     }
 
     @Override
@@ -88,7 +98,7 @@ public void notifyAfterCompleted() throws ServiceNotProvidedException, ModuleSta
 
     @Override
     public String[] requiredModules() {
-        return new String[] {
+        return new String[]{
                 CoreModule.NAME
         };
     }

oap-server/analyzer/gen-ai-analyzer/src/main/java/org/apache/skywalking/oap/analyzer/genai/config/GenAITagKeys.java

@@ -25,4 +25,8 @@ public class GenAITagKeys {
     public static final String INPUT_TOKENS = "gen_ai.usage.input_tokens";
     public static final String OUTPUT_TOKENS = "gen_ai.usage.output_tokens";
     public static final String SERVER_TIME_TO_FIRST_TOKEN = "gen_ai.server.time_to_first_token";
+
+    public static final String ESTIMATED_COST = "gen_ai.estimated.cost";
+
+    public static final String SYSTEM_NAME = "gen_ai.system";
 }