Kwonjae Lee

Kwonjae Lee

SRE, Observability

Weekly - November 15, 2025

- 9 mins

๐Ÿ“š ์ด๋ฒˆ ์ฃผ ์ถ”์ฒœ ์•„ํ‹ฐํด

1. Aurora RDS์˜ ๊ฒฝ์Ÿ ์ƒํƒœ(race condition) ๋ฐœ๊ฒฌ ์‚ฌ๋ก€

์ถœ์ฒ˜: geeknews | ๋‚ ์งœ: 2025-11-15

Hightouch๊ฐ€ AWS Aurora RDS์˜ failover ๊ณผ์ •์—์„œ ๋ฐœ์ƒํ•˜๋Š” race condition์„ ์‹คํ—˜์ ์œผ๋กœ ์ฆ๋ช…ํ•˜๊ณ  AWS๋กœ๋ถ€ํ„ฐ ๊ณต์‹ ํ™•์ธ์„ ๋ฐ›์€ ์‚ฌ๋ก€์ž…๋‹ˆ๋‹ค. ๊ฐœ๋ฐœ์ž๋“ค์€ ํด๋ผ์šฐ๋“œ ์„œ๋น„์Šค์˜ ์•Œ๋ ค์ง€์ง€ ์•Š์€ ๋ฒ„๊ทธ๋ฅผ ์ฒด๊ณ„์ ์œผ๋กœ ๋ฐœ๊ฒฌํ•˜๊ณ  ๊ฒ€์ฆํ•˜๋Š” ๋ฐฉ๋ฒ•์„ ๋ฐฐ์šธ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ๋ถ„์‚ฐ ์‹œ์Šคํ…œ์—์„œ์˜ ์žฅ์•  ์ฒ˜๋ฆฌ์™€ ๊ฒฝ์Ÿ ์ƒํƒœ ๋””๋ฒ„๊น…์— ๋Œ€ํ•œ ์‹ค๋ฌด์ ์ธ ์ธ์‚ฌ์ดํŠธ๋ฅผ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค.

ํ•ต์‹ฌ ํฌ์ธํŠธ:

2. 650GB ๋ฐ์ดํ„ฐ(S3์˜ Delta Lake). Polars vs. DuckDB vs. Daft vs. Spark

์ถœ์ฒ˜: geeknews | ๋‚ ์งœ: 2025-11-15

๊ฐœ๋ฐœ์ž๋“ค์€ 650GB ๊ทœ๋ชจ์˜ S3 Delta Lake ๋ฐ์ดํ„ฐ๋ฅผ ๋‹จ์ผ ๋…ธ๋“œ(32GB ๋ฉ”๋ชจ๋ฆฌ)์—์„œ ์ฒ˜๋ฆฌํ•  ๋•Œ Polars, DuckDB, Daft, Spark์˜ ์‹ค์ œ ์„ฑ๋Šฅ ์ฐจ์ด๋ฅผ ํ•™์Šตํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ๋Œ€์šฉ๋Ÿ‰ ๋ฐ์ดํ„ฐ ์ฒ˜๋ฆฌ ์‹œ ๊ฐ ์—”์ง„์˜ ๋ฉ”๋ชจ๋ฆฌ ํšจ์œจ์„ฑ๊ณผ ์ฒ˜๋ฆฌ ์†๋„๋ฅผ ๋น„๊ตํ•˜์—ฌ ํ”„๋กœ์ ํŠธ์— ์ ํ•ฉํ•œ ๋„๊ตฌ๋ฅผ ์„ ํƒํ•˜๋Š” ๊ธฐ์ค€์„ ์–ป์„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

ํ•ต์‹ฌ ํฌ์ธํŠธ:

3. Continuous profiling for native code: Understanding the what, why, and how

์ถœ์ฒ˜: grafana | ๋‚ ์งœ: 2025-11-14

์ด ๊ธฐ์‚ฌ๋Š” ๋„ค์ดํ‹ฐ๋ธŒ ์ฝ”๋“œ๋ฅผ ์œ„ํ•œ ์—ฐ์† ํ”„๋กœํŒŒ์ผ๋ง์˜ ๊ฐœ๋…๊ณผ ํ•„์š”์„ฑ, ๊ทธ๋ฆฌ๊ณ  ๊ตฌํ˜„ ๋ฐฉ๋ฒ•์„ ์„ค๋ช…ํ•ฉ๋‹ˆ๋‹ค. ๊ฐœ๋ฐœ์ž๋“ค์€ ๋กœ๊ทธ, ๋ฉ”ํŠธ๋ฆญ, ํŠธ๋ ˆ์ด์‹ฑ์— ์ด์–ด ๋„ค ๋ฒˆ์งธ ๊ด€์ฐฐ์„ฑ ๊ธฐ๋‘ฅ์œผ๋กœ ๋ถ€์ƒํ•œ ํ”„๋กœํŒŒ์ผ๋ง์„ ํ†ตํ•ด ์• ํ”Œ๋ฆฌ์ผ€์ด์…˜ ์„ฑ๋Šฅ์„ ์‹ฌ์ธต ๋ถ„์„ํ•˜๋Š” ๋ฐฉ๋ฒ•์„ ํ•™์Šตํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ํŠนํžˆ ํ”„๋กœ๋•์…˜ ํ™˜๊ฒฝ์—์„œ CPU, ๋ฉ”๋ชจ๋ฆฌ ์‚ฌ์šฉ๋Ÿ‰์„ ์‹ค์‹œ๊ฐ„์œผ๋กœ ๋ชจ๋‹ˆํ„ฐ๋งํ•˜์—ฌ ์„ฑ๋Šฅ ๋ณ‘๋ชฉ์ ์„ ์‹๋ณ„ํ•˜๊ณ  ์ตœ์ ํ™”ํ•˜๋Š” ์‹ค๋ฌด์  ๊ธฐ๋ฒ•์„ ์ตํž ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

ํ•ต์‹ฌ ํฌ์ธํŠธ:

4. OpenTelemetry eBPF Instrumentation Marks the First Release

์ถœ์ฒ˜: opentelemetry | ๋‚ ์งœ: 2025-11-03

Grafana Labs, Splunk, Coralogix, Odigos ๋“ฑ ์ฃผ์š” ๊ธฐ์—…๋“ค๊ณผ ์ปค๋ฎค๋‹ˆํ‹ฐ์˜ ํ˜‘๋ ฅ์œผ๋กœ OpenTelemetry eBPF Instrumentation์˜ ์ฒซ ๋ฒˆ์งธ ์•ŒํŒŒ ๋ฆด๋ฆฌ์Šค๊ฐ€ ๋ฐœํ‘œ๋˜์—ˆ์Šต๋‹ˆ๋‹ค. ์ด ๋„๊ตฌ๋Š” ์• ํ”Œ๋ฆฌ์ผ€์ด์…˜ ์ฝ”๋“œ ์ˆ˜์ • ์—†์ด ์ž๋™์œผ๋กœ ํ…”๋ ˆ๋ฉ”ํŠธ๋ฆฌ ๋ฐ์ดํ„ฐ๋ฅผ ์ˆ˜์ง‘ํ•  ์ˆ˜ ์žˆ๋Š” ํ˜์‹ ์ ์ธ ๊ด€์ฐฐ๊ฐ€๋Šฅ์„ฑ ์†”๋ฃจ์…˜์„ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค. ๊ฐœ๋ฐœ์ž๋“ค์€ ๊ธฐ์กด ์• ํ”Œ๋ฆฌ์ผ€์ด์…˜์— ์ตœ์†Œํ•œ์˜ ์นจ์ž…์œผ๋กœ ์„ฑ๋Šฅ ๋ชจ๋‹ˆํ„ฐ๋ง๊ณผ ๋ถ„์‚ฐ ์ถ”์ ์„ ๊ตฌํ˜„ํ•  ์ˆ˜ ์žˆ๊ฒŒ ๋˜์—ˆ์Šต๋‹ˆ๋‹ค.

ํ•ต์‹ฌ ํฌ์ธํŠธ:

5. Understand, diagnose, and optimize SQL queries: Introducing Grafana Cloud Database Observability

์ถœ์ฒ˜: grafana | ๋‚ ์งœ: 2025-11-13

Grafana Cloud Database Observability๋Š” ์• ํ”Œ๋ฆฌ์ผ€์ด์…˜ ์„ฑ๋Šฅ ๋ฌธ์ œ์˜ ์ฃผ์š” ์›์ธ์ธ ๋ฐ์ดํ„ฐ๋ฒ ์ด์Šค ์ฟผ๋ฆฌ๋ฅผ ์ดํ•ดํ•˜๊ณ  ์ง„๋‹จํ•˜๋ฉฐ ์ตœ์ ํ™”ํ•  ์ˆ˜ ์žˆ๋Š” ๋„๊ตฌ๋ฅผ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค. ๊ฐœ๋ฐœ์ž๋“ค์€ ๋А๋ฆฌ๊ฑฐ๋‚˜ ๋น„ํšจ์œจ์ ์ธ ๋ฐ์ดํ„ฐ๋ฒ ์ด์Šค ์ฟผ๋ฆฌ๋ฅผ ์‹๋ณ„ํ•˜๊ณ  ํ•ด๊ฒฐํ•˜์—ฌ ์ „์ฒด ์• ํ”Œ๋ฆฌ์ผ€์ด์…˜ ์„ฑ๋Šฅ์„ ํ–ฅ์ƒ์‹œํ‚ฌ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ์ด๋ฅผ ํ†ตํ•ด ๋ฐ์ดํ„ฐ๋ฒ ์ด์Šค ์„ฑ๋Šฅ ๋ชจ๋‹ˆํ„ฐ๋ง๊ณผ ์ตœ์ ํ™”์— ๋Œ€ํ•œ ์‹ค์งˆ์ ์ธ ์ธ์‚ฌ์ดํŠธ๋ฅผ ์–ป์„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

ํ•ต์‹ฌ ํฌ์ธํŠธ:

6. Go์˜ 16๋ฒˆ์งธ ์ƒ์ผ

์ถœ์ฒ˜: geeknews | ๋‚ ์งœ: 2025-11-15

Go ์–ธ์–ด์˜ 16์ฃผ๋…„์„ ๋งž์•„ ์ตœ๊ทผ 1๋…„๊ฐ„ ์ด๋ฃจ์–ด์ง„ ์ฃผ์š” ๊ธฐ์ˆ  ๋ฐœ์ „์„ ์‚ดํŽด๋ณผ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. Go 1.24์™€ 1.25 ๋ฒ„์ „์—์„œ๋Š” ํ…Œ์ŠคํŠธ, ๋ณด์•ˆ, ์„ฑ๋Šฅ ์˜์—ญ์—์„œ ๋Œ€ํญ์ ์ธ ๊ฐœ์„ ์ด ์ด๋ฃจ์–ด์กŒ์œผ๋ฉฐ, synctest์™€ container-aware scheduling ๋“ฑ์˜ ์ƒˆ๋กœ์šด ๊ธฐ๋Šฅ์ด ๋„์ž…๋˜์—ˆ์Šต๋‹ˆ๋‹ค. ๊ฐœ๋ฐœ์ž๋“ค์€ ํ–ฅํ›„ Go ์ƒํƒœ๊ณ„์˜ ๋ฐœ์ „ ๋ฐฉํ–ฅ๊ณผ ํ™œ์šฉ ๊ฐ€๋Šฅํ•œ ์ƒˆ๋กœ์šด ๊ธฐ๋Šฅ๋“ค์— ๋Œ€ํ•ด ํŒŒ์•…ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

ํ•ต์‹ฌ ํฌ์ธํŠธ:

7. Performance testing best practices: How to prepare for peak demand with Grafana Cloud k6

์ถœ์ฒ˜: grafana | ๋‚ ์งœ: 2025-11-12

์ด ๋ฌธ์„œ๋Š” ๋ธ”๋ž™ ํ”„๋ผ์ด๋ฐ์ด, ์ œํ’ˆ ์ถœ์‹œ, ๋Œ€๊ทœ๋ชจ ์„ธ์ผ ๋“ฑ ๋†’์€ ๊ณ ๊ฐ ํ™œ๋™ ๊ธฐ๊ฐ„์— ๋Œ€๋น„ํ•œ ์„ฑ๋Šฅ ํ…Œ์ŠคํŠธ ๋ชจ๋ฒ” ์‚ฌ๋ก€๋ฅผ ๋‹ค๋ฃน๋‹ˆ๋‹ค. ๊ฐœ๋ฐœ์ž๋“ค์€ Grafana Cloud k6๋ฅผ ํ™œ์šฉํ•˜์—ฌ ํ”ผํฌ ์ˆ˜์š” ์ƒํ™ฉ์—์„œ ์†Œํ”„ํŠธ์›จ์–ด์™€ ์ธํ”„๋ผ๊ฐ€ ์•ˆ์ •์ ์œผ๋กœ ์ž‘๋™ํ•˜๋„๋ก ๋ณด์žฅํ•˜๋Š” ๋ฐฉ๋ฒ•์„ ํ•™์Šตํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค. ์˜ˆ์ƒ์น˜ ๋ชปํ•œ ํŠธ๋ž˜ํ”ฝ ๊ธ‰์ฆ์œผ๋กœ ์ธํ•œ ์‹œ์Šคํ…œ ์žฅ์• ๋ฅผ ๋ฏธ์—ฐ์— ๋ฐฉ์ง€ํ•˜๋Š” ์‹ค๋ฌด์ ์ธ ์ ‘๊ทผ๋ฒ•์„ ์ œ๊ณตํ•ฉ๋‹ˆ๋‹ค.

ํ•ต์‹ฌ ํฌ์ธํŠธ:


๐Ÿ“š This Weekโ€™s Picks

1. Aurora RDS์˜ ๊ฒฝ์Ÿ ์ƒํƒœ(race condition) ๋ฐœ๊ฒฌ ์‚ฌ๋ก€

Source: geeknews | Date: 2025-11-15

This case study shows how Hightouch experimentally discovered and proved a race condition bug in AWS Aurora RDSโ€™s failover process, ultimately receiving official confirmation from AWS. Developers can learn systematic approaches to identifying and validating unknown bugs in cloud services. It provides practical insights into failure handling and race condition debugging in distributed systems.

Key Points:

2. 650GB ๋ฐ์ดํ„ฐ(S3์˜ Delta Lake). Polars vs. DuckDB vs. Daft vs. Spark

Source: geeknews | Date: 2025-11-15

Developers can learn about real-world performance differences between Polars, DuckDB, Daft, and Spark when processing 650GB Delta Lake data from S3 on a single node with 32GB memory. This comparison provides practical insights into memory efficiency and processing speed of each engine, helping developers choose the right tool for their large-scale data processing projects.

Key Points:

3. Continuous profiling for native code: Understanding the what, why, and how

Source: grafana | Date: 2025-11-14

This article explains the concept, necessity, and implementation methods of continuous profiling for native code. Developers can learn how to perform in-depth application performance analysis through profiling, which has emerged as the fourth pillar of observability alongside logs, metrics, and tracing. They will gain practical techniques for identifying performance bottlenecks and optimizing applications by monitoring CPU and memory usage in real-time within production environments.

Key Points:

4. OpenTelemetry eBPF Instrumentation Marks the First Release

Source: opentelemetry | Date: 2025-11-03

The first alpha release of OpenTelemetry eBPF Instrumentation has been announced through significant collaboration between Grafana Labs, Splunk, Coralogix, Odigos, and community members. This tool provides an innovative observability solution that can automatically collect telemetry data without modifying application code. Developers can now implement performance monitoring and distributed tracing with minimal intrusion to existing applications.

Key Points:

5. Understand, diagnose, and optimize SQL queries: Introducing Grafana Cloud Database Observability

Source: grafana | Date: 2025-11-13

Grafana Cloud Database Observability provides tools to understand, diagnose, and optimize database queries, which are often the primary cause of application performance issues. Developers can identify and resolve slow or inefficient database queries to improve overall application performance. This enables practical insights into database performance monitoring and optimization strategies.

Key Points:

6. Go์˜ 16๋ฒˆ์งธ ์ƒ์ผ

Source: geeknews | Date: 2025-11-15

This article covers Go languageโ€™s major technical advancements over the past year in celebration of its 16th anniversary. Go versions 1.24 and 1.25 introduce significant improvements across testing, security, and performance areas, including new features like synctest and container-aware scheduling. Developers can learn about the future direction of the Go ecosystem and new capabilities they can leverage in their projects.

Key Points:

7. Performance testing best practices: How to prepare for peak demand with Grafana Cloud k6

Source: grafana | Date: 2025-11-12

This article covers performance testing best practices for preparing systems during high customer activity periods like Black Friday, product launches, and major sales events. Developers will learn how to use Grafana Cloud k6 to ensure their software and infrastructure can handle peak demand reliably. It provides practical approaches to prevent system failures caused by unexpected traffic spikes.

Key Points:


๐Ÿ”— Sources

Articles curated from various tech blogs and communities including SRE Weekly, GeekNews, OpenTelemetry, Grafana, and more.


์•„ํ‹ฐํด ์ œ์•ˆ์ด ์žˆ์œผ์‹œ๋ฉด ์ด๋ฉ”์ผ๋กœ ์—ฐ๋ฝ์ฃผ์‹œ๊ฑฐ๋‚˜ ๋Œ“๊ธ€์„ ๋‚จ๊ฒจ์ฃผ์„ธ์š”!

Have an article suggestion? Feel free to reach out via email or leave a comment below!

Kwonjae Lee

Kwonjae Lee

SRE, Observability

comments powered by Disqus