lmolkova

private string GetFinishReason(ChatFinishReason finishReason) => finishReason switch
{
    ChatFinishReason.Stop => "stop",
    ChatFinishReason.Length => "length",
    ChatFinishReason.ToolCalls => "tool_calls",
    // ...
    _ => finishReason.ToString(),
};

private void LogChoice(ChatCompletion message)

lmolkova

import json
import os
import time
from typing import override
from opentelemetry.instrumentation.httpx import HTTPXClientInstrumentor
#HTTPXClientInstrumentor().instrument()

from openai import AssistantEventHandler
from openai.types.beta.threads.runs import RunStep

lmolkova

public class Main {
  
    public static void main(String[] args) throws IOException {
        var logProvider = SdkLoggerProvider.builder()
          .addLogRecordProcessor(SimpleLogRecordProcessor.create(OtlpGrpcLogRecordExporter.getDefault())).build();
        EventLoggerProvider eventLoggerProvider = SdkEventLoggerProvider.create(logProvider);

        var logger = eventLoggerProvider.get("test-logger");
        var event = new GenAIChoice()
                .setId(1)

lmolkova

let get_role = (event_name:string) { 
  iff(event_name == "gen_ai.choice", "assistant", split(event_name, ".")[1])
};
let gen_ai_spans = dependencies
| where isnotnull(customDimensions["gen_ai.system"])
| project 
    operation_ParentId,
    operation_Id,
    id,
    duration,

lmolkova

# set OPENAI_API_KEY
# install otel-api and otel-sdk from main
# install opentelemetry-instrumentation-openai from https://github.com/lmolkova/opentelemetry-python-contrib/tree/openai-events

from time import time_ns
from openai import OpenAI

from typing import Optional
from opentelemetry import trace, _logs, _events
import opentelemetry.sdk

lmolkova

Title	Link	# of reactions	Reactions
Capture request and response bodies	open-telemetry/semantic-conventions#857	29	{'+1': 28, '-1': 1, 'laugh': 0, 'hooray': 0, 'confused': 0, 'heart': 0, 'rocket': 0, 'eyes': 0}
Introduce semantic conventions for modern AI (LLMs, vector databases, etc.)	open-telemetry/semantic-conventions#327	11	{'+1': 0, '-1': 0, 'laugh': 0, 'hooray': 8, 'confused': 0, 'heart': 3, 'rocket': 0, 'eyes': 0}
Improve GraphQL semantic conventions	open-telemetry/semantic-conventions#182	8	{'+1': 1, '-1': 0, 'laugh': 0, 'hooray': 0, 'confused': 0, 'heart': 0, 'rocket': 6, 'eyes': 1}
Discuss ECS' field reuse concept for OTel semantic attributes	open-telemetry/semantic-conventions#339	7	{'+1': 7, '-1': 0, 'laugh': 0, 'hooray': 0, 'confused': 0, 'heart': 0, 'rocket': 0, 'eyes': 0}
Define semantic conven

lmolkova

Prereqs

• opentelemetry-sdk, azure-monitor-opentelemetry-exporter
• set APPLICATIONINSIGHTS_CONNECTION_STRING (or remove AzMon exporter)

Result

lmolkova

What's Micrometer story for library owners

• Which one to use Otel or Micrometer. Why?
• Semantic conventions - how users can pick which ones they want? will libs need to provide all possible options?
• What's the story in other languages/ecosystems? I need my python messaging consumer to be consistent with my java messaging producer.
• Which libraries beyond Spring are all-in on Micrometer?
• How do you prioritize features? What if something is missing? (E.g. structured events on spans, adding link after span creation, etc)
  • is Micrometer targeting feature-parity with OTel specification? Will micrometer introduce profling API?
• How big is the community? Are there community calls? Who makes decisions?

lmolkova

Load testing event fan-out scenario with Azure Event Hubs SDK for Java

• Scenario
• Code
• Configuration
  • Environment
• Results
  • Event Hubs SDK versions
  • Available CPU and memory
• Fan-out factor

lmolkova

Checking ServiceBusProcessorClient performance (via Spring binder)

Setup

Tested processor performance in the following setup:

Sender sends up to 4k messages per second in batches. Message size is 12 bytes.