Twitch 宣佈退出韓國市場

Twitch 宣佈 2024/02/27 (星期二) 退出韓國市場:「An Update on Twitch in Korea」。日期不知道是怎麼選的,可能跟某些合約有關?

Twitch 目前的公告會有繁體中文,也可以看這份:「Twitch 韓國現況更新」。

另外今天早上找了一下,Hacker News 也有討論了:「An update on Twitch in Korea (」。


Ultimately, the cost to operate Twitch in Korea is prohibitively expensive and we have spent significant effort working to reduce these costs so that we could find a way for the Twitch business to remain in Korea.

這邊提到的包括了 p2p model 以及降到 720p,但即使如此網路費用 (應該就是頻寬費用) 是其他區域的十倍以上:

First, we experimented with a peer-to-peer model for source quality. Then, we adjusted source quality to a maximum of 720p. While we have lowered costs from these efforts, our network fees in Korea are still 10 times more expensive than in most other countries. Twitch has been operating in Korea at a significant loss, and unfortunately there is no pathway forward for our business to run more sustainably in that country.

Cloudflare 這邊,2016 年還叫做 CloudFlare 的時候也有抱怨過:「CloudFlare 對 HiNet 成本的抱怨 (還有其他 ISP...)」。

當年是這樣寫 HiNetKT,成本大約是歐美區的 15 倍:

Two Asian locations stand out as being especially expensive: Seoul and Taipei. In these markets, with powerful incumbents (Korea Telecom and HiNet), transit costs 15x as much as in Europe or North America, or 150 units.


South Korea is perhaps the only country in the world where bandwidth costs are going up. This may be driven by new regulations from the Ministry of Science, ICT and Future Planning, which mandate the commercial terms of domestic interconnection, based on predetermined “Tiers” of participating networks. This is contrary to the model in most parts of the world, where networks self-regulate, and often peer without settlement. The government even prescribes the rate at which prices should decrease per year (-7.5%), which is significantly slower than the annual drop in unit bandwidth costs elsewhere in the world. We are only able to peer 2% of our traffic in South Korea.

不過不確定現在的情況,2016 年的 CloudFlare 跟 2023 年的 Cloudflare 已經差了七年了...

最近 AV1 的支援度

HN 上「AV1 video codec gains broader hardware support (」這篇在說 AV1 的支援度變得更好了,原文不長,在「The AV1 video codec gains broader hardware support」這邊。

Can I Use 上的 AV1 video format 可以看的比較清楚:

不過在瀏覽器上離直接取代掉其他的 video codec 還早,但算是個起頭,至少 iPhone 15 Pro 與 iPhone 15 Pro Max 上的 Safari 支援了,接下來就是看桌機的 Edge 什麼時候才又想到要把 AV1 開回來:

Edge has stopped supporting AV1 completely at some point prior to version 116 (additional information required).

擋 YouTube 短影音的設定

短影音類的影片因為沒有知識量 (沒有 reference 可以確認正確性),我完全不會看... 但 YouTube 上一般的影音我會翻,所以就會冒出這個需求了。

YouTube 的短影音有幾個地方會出現 (補充一下,我這邊是用英文版介面):

  • 首頁的左邊,會有一個 Shorts 的連結可以點進 Shorts 看。
  • 首頁的推薦裡面也會有 Shorts 的 section。


這邊要注意的是,後者除了擋掉 Shorts 以外,還會擋掉各種 YouTube 的推銷 (像是電影之類的),這個也是我要擋的,所以我這邊直接用了 ytd-rich-section-renderer 這個元素來擋。

再來是各種穿插在頁面裡面的 Shorts 內容,像是首頁、訂閱頁與搜尋結果頁,這些就要找出對應的元素來擋:[href*="/shorts/"])

另外一個跟短影音無關,但還是很影響專注度的是,YouTube 的搜尋結果會給你一堆很干擾結果的推薦,像是「People also watched」、「For you」、「Previously watched」以及「From related searches」,也可以設定擋掉: ytd-shelf-renderer[thumbnail-style]


Raspberry Pi 5 拿掉硬體的 H.264 encoding

HN 上看到「Raspberry Pi 5 has no hardware video encoding and only HEVC decoding (」,原文指到 Gordon Hollingworth 的回覆這邊,可以看到這是上個月的消息了。

Raspberry Pi 一直都有硬體 H.264 encoding 的能力,不過這個在 Raspberry Pi 5 上被拿掉了,所以得用軟體壓。

官方有提到 Raspberry Pi 5 的 CPU 因為比之前快很多,單顆就有辦法做到 1080p60 (而 RPi5 有 4 CPU),所以除了 power consumption 以外應該不是大問題:

Obviously, the bad thing is the power consumption, but actually it only takes around 1 processor to encode 1080p60 with our default settings (which is still better quality than the PI 4 hardware encoder).

不過 HN 上猜是授權費用之類的問題,我在想是不是新的晶片組的 encoding license 都綁在一起,H.264 + H.265 得一包一起買,而 H.265 的授權費是眾所皆知的貴...

當 desktop 應該是還好,但就心裡有個底...

下載 YouTube 影片的技術限制與繞過方法

Hacker News 上看到這篇「How They Bypass YouTube Video Download Throttling」在講 YouTube 防止下載的各種方式。

透過 API 拿到的 URL 直接抓很慢,大約 40-70KB/sec:

However, attempting to download from this URL leads to really slow download:

The speed is always limited to around 40-70kB/s.

這邊需要一個 javascript 環境計算出 n,帶入後續的 request 以「證明」你是官方的網頁 client:

Since mid-2021, YouTube has included the query parameter n in the majority of file URLs. This parameter needs to be transformed using a JavaScript algorithm located in the file base.js, which is distributed with the web page. YouTube utilizes this parameter as a challenge to verify that the download originates from an “official” client. If the challenge is not resolved and n is not transformed correctly, YouTube will silently apply throttling to the video download.

The JavaScript algorithm is obfuscated and changes frequently, so it’s not practical to attempt reverse engineering to understand it. The solution is simply to download the JavaScript file, extract the algorithm code, and execute it by passing the n parameter to it. The following code accomplishes this.

但即使算出 n,也還是會限速,可以看到作者策出來大約是 4MB/sec,雖然比以前快很多了,但還是看得出來有限速。這主要是避免 client 端過度 buffer 浪費頻寬:

With this new URL containing the correctly transformed n parameter, the next step is to download the video. However, YouTube still enforces a throttling rule. This rule imposes a variable download speed limit based on the size and length of the video, aiming to provide a download time that’s approximately half the duration of the video. This aligns with the streaming nature of videos. It would be a massive waste of bandwidth for YouTube to always provide the media file as quickly as possible.

接下來的方式就是利用 Range 拆成很多個 HTTP request 打,這樣因為 buffering algorithm 在開始限速前會先全速塞資料給你,就可以用這點避開限速的問題了。

把多的 request 與處理時間都算進去後,整體大約可以到 50-70MB/sec,算是可以接受的下載速度了:

However, the average speeds typically ranged between 50-70 MB/s or 400-560 Mb/s, which is still pretty fast.

後面有一些合併處理的指令 (因為 YouTube 會把影與音分離成兩個檔案),就不是重點了...

雲端上面的 GPU 資源費用,以及地端的 GPU 決策圖

Hacker News 上面看到「Cloud GPU Resources and Pricing (」這篇,原網頁是「Cloud GPUs - The Full Stack」,裡面有些有用的資源可以拉出來獨立看。

雲端的選擇上,因為 H100 看起來還沒普及,所以用上一代的 A100 (80GB) 來看,可以看到大的雲端跟其他家的差異還是蠻大的:

不過這邊好像沒把 放進來。

地端的資訊主要是直接購買顯示卡時的選擇,可以看到如果除了各系列的旗艦卡外 (4090 & 3090 & 2080),3060 是一張會在考慮到「便宜」而上榜的卡,應該是因為他是一張入門價位的顯卡,卻有 12GB VRAM 的關係:

在接下來七月要推出的 4060 會出 16GB VRAM 版本,應該會取代現在 3060 12GB VRAM 的地位...

llama.cpp 開始支援 GPU 了

前陣子因為重灌桌機,所以在重建許多環境... 其中一個就是 llama.cpp,連到專案頁面上時意外發現這兩個新的 feature:

OpenBLAS support
cuBLAS and CLBlast support

這代表可以用 GPU 加速了,所以就照著說明試著編一個版本測試。

編好後就跑了 7B 的 model,看起來快不少,然後改跑 13B 的 model,也可以把完整 40 個 layer 都丟進 3060 (12GB 版本) 的 GPU 上:

./main -m models/13B/ggml-model-q4_0.bin -p "Building a website can be done in 10 simple steps:" -n 512 -ngl 40

從 log 可以看到 40 layers 到都 GPU 上面,吃了 7.5GB 左右:

llama.cpp: loading model from models/13B/ggml-model-q4_0.bin
llama_model_load_internal: format     = ggjt v2 (latest)
llama_model_load_internal: n_vocab    = 32000
llama_model_load_internal: n_ctx      = 512
llama_model_load_internal: n_embd     = 5120
llama_model_load_internal: n_mult     = 256
llama_model_load_internal: n_head     = 40
llama_model_load_internal: n_layer    = 40
llama_model_load_internal: n_rot      = 128
llama_model_load_internal: ftype      = 2 (mostly Q4_0)
llama_model_load_internal: n_ff       = 13824
llama_model_load_internal: n_parts    = 1
llama_model_load_internal: model size = 13B
llama_model_load_internal: ggml ctx size =  90.75 KB
llama_model_load_internal: mem required  = 9807.48 MB (+ 1608.00 MB per state)
llama_model_load_internal: [cublas] offloading 40 layers to GPU
llama_model_load_internal: [cublas] total VRAM used: 7562 MB
llama_init_from_file: kv self size  =  400.00 MB

30B 的 model 我也試著丟上去跑,但只能丟 28 layers 上去 (全部是 60 layers),再多 GPU 的記憶體就撐不住了。

但能用 GPU 算是一個很大的進展,現在這版只快了一半的時間,不知道後面還有沒有 tune 的空間...

另外一則跟 Prime Video 有關的有趣留言

Hacker News 上的「Even Amazon can't make sense of serverless or microservices (」這邊看到 DHH 抓到機會剛好戳了 AWS 的費用問題,不過讓我注意到的不是 DHH 的文章,而是在 Hacker News 上的留言 35823366,自稱是當時設計這個 serverless 架構的人出來解釋當時的確有壓力測試過,看起來也還 OK:

I actually designed the original serverless system (a few years back when I was still at Prime Video), and yes we did and it did sort of look like it could work until it didn't. Obviously wasn't the right solution for the scale we had in mind (or rather the type of problem we were working on) but it's sad to see the mistake of one team be used to justify shitting on serverless as a general solution.

但這是全新帳號的發言,拿這個 id 去找是可以翻到 TimboKZ 這個 GitHub 帳號,然後一路找也可以看到 Timur KuzhagaliyevLinkedIn 帳號,這個帳號裡面有提到 2019 到 2020 的時候在 Amazon 裡面當 SDE:

Worked on various computer vision projects as a part of Video Quality Analysis team at Prime Video.

是都符合沒錯,但還是不能完全肯定,目前看八卦的心態還是比較多... CTO 解釋 Prime Video 的那篇文章

先前在「Amazon Prime Video 捨棄 AWS Step Functions 回頭用 EC2 與 ECS 省錢的文章」這篇裡面有提到 Prime Video 團隊寫的文章,有點負面 PR 的感覺。

而讓人意外的是 的 CTO Werner Vogels 也寫了一篇文章提到這件事情:「Monoliths are not dinosaurs」。

從標題可以看到他是針對 monolith 架構以及其他的架構在討論,而不是在講 AWS 成本問題 (這反而是原本 Prime Video 那篇文章在標題宣傳的),另外也講了一些管理面上的東西,也就是文章裡面的副標題,以及兩段粗體字:

Building evolvable software systems is a strategy, not a religion. And revisiting your architectures with an open mind is a must.

If you hire the best engineers, you should trust them to make the best decisions.

there is not one architectural pattern to rule them all.

至於算不算是提油... 就見仁見智了。

Amazon Prime Video 捨棄 AWS Step Functions 回頭用 EC2 與 ECS 省錢的文章

昨天在 Hacker News 上熱烈討論的文章,是一篇三月就放出來,但昨天被丟上來意外的熱烈討論,在講 Amazon Prime Video 的團隊改寫程式,把 AWS Step Functions 拔掉,並且回頭用 EC2ECS 而省下大量 AWS 費用的文章討論:「Scaling up the Prime Video audio/video monitoring service and reducing costs (」,原文在「Scaling up the Prime Video audio/video monitoring service and reducing costs by 90%」,Internet Archive 的備份Archive Today 的備份

先看文章的部分,裡面提到了他們用 AWS Step Functions,但意外的貴:

The initial version of our service consisted of distributed components that were orchestrated by AWS Step Functions. The two most expensive operations in terms of cost were the orchestration workflow and when data passed between distributed components.

然後改寫程式把所有東西都放在單一 process 裡面跑就好,用標準的 EC2 或是 ECS 就可以 scale 很好,而且也省錢:

To address this, we moved all components into a single process to keep the data transfer within the process memory, which also simplified the orchestration logic. Because we compiled all the operations into a single process, we could rely on scalable Amazon Elastic Compute Cloud (Amazon EC2) and Amazon Elastic Container Service (Amazon ECS) instances for the deployment.

可以看出起因是一開始設計時的 overdesign,把可以簡單處理的東西拆開,另外加上雲端在這塊收費特別貴而導致成本爆增... 這件事情偶而會發生,尤其是比較新的東西會沒注意到成本,通常在上線發現不太對的時候就會安排 refactor 掉。

但如果是 Amazon 自家集團的其他團隊出來抱怨,就有很棒的 PR 效果了,所以 Hacker News 上就看到有人在猜可能過不久後文章就會不見 XD (但文章紅了以後應該就不會不見 XD):

My word. I'm sort of gob smacked this article exists.

I know there are nuances in the article, but my first impression was it's saying "we went back to basics and stopped using needless expensive AWS stuff that caused us to completely over architect our application and the results were much better". Which is good lesson, and a good story, but there's a kind of irony it's come from an internal Amazon team. As another poster commented, I wouldn't be surprised if it's taken down at some point.

很政治不正確的文章 XD

以之前的經驗來說,AWS 上類似的東西還包括了 NAT Gateway,這東西只適合在有強資安需求 (像是法規要求),而且需要連外的流量很少的時候適合。

NAT Gateway 在新加坡 ap-southeast-1 要 $0.059/hr (美金,所以大約是 $42.48/mo),以及 US$0.045/GB 的處理費用,所以假設你每天只有 100GB (平均 10Mbps),就等於是 3TB/mo,要 $135/mo。這樣整包就 $172.48/mo 了。

如果讓 EC2 機器直接連去 internet 抓資料的話,這些費用就是 $0,你只要付無論是有 NAT Gateway 或是沒有 NAT Gateway 的 outbound traffic 費用部分 (大多是各種 TCP/TLS/HTTP header)。

比較省成本的解法是用 security group 對 outbound traffic 開放特定的流量來解。

另外一種方式還是 NAT,但是是自己架設 HA 的 NAT service,像是 2015 年的文章「The Right Way to set up NAT in EC2」提到的方法。

這個方法以現在的機種來說,兩台 t4g.nano 的機器加上 EBS 不到 $10/mo,唯一要注意的應該是網路頻寬雖然可以 burst 到 5Gbps,但他的網路頻寬是 credit 機制,當 credit 用完的時候 t4g.nano 記得是剩下 100Mbps 左右?不過真的有這個量的時候機器也可以往上開大一點...

另外還有很多「好用」的雲端服務,但看到帳單後就變得「不好用」的雲端服務... 在用之前先算一下成本就會發現了。