---
tags: 2021-organize, organize, data squad, 2021-data-squad
---

🔙 Back to [歷年 PyCon TW Organizing 共筆](/ryPr7SFyP/%2FHM5mHCFKQCu7-W5ea8ITcw%3Fview)
🔙 Back to [PyCon TW 2021 Organizing 共筆](/Wb9vQrfJQk-5tPoPR23hwA)

## Meeting Minute Template
:::info
- **Location:** google meet
- **Link**:
- **Date:** 15:30 June 16, 2021 (TST)
- **Agenda:**
1. 沒有 Spark 之前，大家用什麼 framework? 他有什麼缺點？
2. Spark 的 pros and cons 是什麼？歡迎分享自己踩坑經驗
3. 有其他 framework 能取代 Spark 嗎？
- **Participants:**
1. david Jr
2. tai
3. hane
4. juihsiang
5. shirley
6. grimmer
7. hyw
- **Online Participants:**
- **Minutes Taker:**
:::

### 讀書會 Note

- google 發 GFS paper， hadoop 則是實做
- hadoop file system (hdfs) is implemented based on GFS theory
- pros / cons
- pros
- spark 迭代比較快， hadoop 要先寫入硬碟
- tolerlance is better
- [name=shirley] I referred to https://www.itread01.com/content/1550363233.html
- [google beam](https://beam.apache.org/) 可以接 spark and flink ([apache flink](https://flink.apache.org/))
- comparing to "spark mini batch", flink has better 即時性，也許更適合客服機器人

### A.O.B. （臨時動議）