HN user

flybarrel

44 karma

Desperately hope the world sucks less

Posts0
Comments28
View on HN
No posts found.

In the digital age, almost impossible. Documentation, process automation, and knowledge transfer mitigate this risk.

Are you sure you really work at AWS? Dude, how often do you realize the documentation is not sufficient and you have to dig through people's brain to actually get the nuances, that is, lucky enough they are still around? lolll

There's a thing called "local laws and regulations" that you need to comply with to be able to operate in China. It's plain and simple - without this level of limitation, once the model is viral it will be on the radar and then censorship will apply anyway. May as well implement that from the beginning. So I don't believe CCP is actively "involved" in this, but rather the laws impacted the behavior of the company.

Microsoft apply censorship to Bing search results in China. It doesn't mean they are controlled by CCP. They just got impacted by law and they want to keep operate in China.

Back in the early stage of social media, US companies had the choice to operate in China as long as they comply with the censorship and local laws. Had they chosen not to quit China market at the point, they would have been probably huge in China holding major access over Chinese users too. (How would Chinese government react to that is something we never get to see now...)

I keep seeing argument regarding "China bans social medias from other countries". It's not an outright ban saying that "Facebook cannot operate in China", but more like "Comply with the censorship rules or you cannot operate in China". It's not targeting "ownership" or "nation states". e.g. Google chose to leave, while Microsoft continues to operate Bing in China.

| I have faith that UW leadership will make it right in the end.

Please do not plan based on anyone's goodwill only, especially in this one you have such a high stake. I'd at least consult a lawyer at this point, even if it means you don't take any legal action.

$7.25 minimum wage per hour at fed level. 40 hour per week and 50 working week you have 14,500 USD, or 105K RMB roughly based on exchange rate today.

105K RMB yearly is close to or above average even in the above cities.

for some state the minimum wage can be $15 or higher, this is wayyyy better than China.

You are likely right the code is not audited, or deployment is not verified. This is a design-level review.

You are inaccurate on "they only interviewed engineers". There are document reviews as well which covers design and architectures.

Source of the report on methodology shared such information.

Yes throughput for EBS is also relevant to the instance type. It is all published here: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ebs-opti...

EBS connects to EC2 via a separate pipeline, different from the EC2 instance Networking bandwidth. This is true for all Nitro instances. EFS / FSx connects to EC2 via Networking bandwidth. So you should refer to that if you are looking for the bandwidth information.

Depends on what volume types you are using. For random access - with io2 you can get 256K IOPS per volume, and if you do RAID0, on the largest instance you can get 400K IOPS. Directly-attached vs over-the-network storage can be fairly different, but going back to a spinning drive is not a fair statement.

Not sure what you are talking about. Garman was in product since joined AWS till 2020, then to Sales / marketing.

Exactly. It's the risk management system that a company implements. Tech companies also serve customers who are in competition with each other. No company is above this. It's all about how you handle these tricky situations. I have yet to see concrete evidences saying that the internal firewall was breached. I'll wait and see.

1MT will take out the infrastructure to make the data not Available. However the data is still in the 3rd datacenter, making it still accessible therefore no compromising on Durable but yes we don't need the cat pictures when that is close to home :)

Newbie question - What happens after when an LLM training job experience a hardware failure? I don't suppose you lose all the training progress do you? Then the pain is mostly in the diagnostic of the problem and getting the cluster running again, but no need to worry about data loss right?