seedless-sensat

It was only 600 words. I understand there is a push against AI bloated slop here, but I don't think we should discourage good prose.

I just wanted to share that I find Seth's daily blog consistently excellent.

What is an ATS? This blog doesn't define it

I am not American, but I think you are unfairly dismissing the massive benefits GLP-1s have for people who struggle to maintain a healthy diet. It really feels like a miracle drug.

> It's insane to me that so many people need these to get off the processed foods killing them in the US.

Your comparison of your friends in Germany vs "insanity" in the US doesn't feel relevant

From the page:

> API health is inferred by analyzing aggregated and anonymized telemetry from across the Datadog customer base.

Adding an alternate data point, I was a heavy Linux desktop user, and had an adjustment period when my workplace gave me a Mac 10 years ago. Yes there are random differences. However, now I wouldn't look back for my personal compute needs.

Came to make the same comment. I got through like 12 paragraphs, and it still hadn't explained what a Semantic Layer is, so I gave up

The landing great lever is shaped like a wheel as a design affordance. It would be VERY hard to confuse

I think you jumped to the wrong conclusion. Another news source suggests proper groundbreaking 10C charging:for a standard EV

> the BYD’s pile supports the 10C charging. It can charge 400 km in 5 minutes. It is two kilometers in one second! During the live test, this station reached the 1 MW level of power in 10 seconds (while charging Han L EV and Tang L EV). The car’s charging time from 7% to 50% was just 4.5 minutes.

Looks interesting. Just wondering, how did you decide to use NATS as the transport, instead of Kafka?

In the article, they are using a PPS output into a GPIO IRQ. I don't think they're using serial/NMEA for the timestamping

Obviously this is subjective, I just wanted to say that I personally found it's production to be incredibly beautiful.

Why are you protecting Google's internal architecture onto to AWS? Your Google mental model is not correct here

I think this con is very real:

> Related tables and indexes are not necessarily stored together, meaning typical operations such as joins and evaluating foreign keys or even simple index lookups might incur an excessive number of internal network hops. The relatively strong transactional guarantees that involve additional locks and coordination can also become a drag on performance.

You handwaved this away saying you can just store an entire table on a single node, but that defeats many of the benefits of these sharded SQL databases.

Edit: Also, before attacking the author's biases, it seems fair to disclose you appear to work at Yugabyte

What is RAG? That's hard to search for

I am looking at this while sitting on Amtrak. It is about 14mi behind my current position, but still very cool!

How does firing the CEO help in your scenario? Now they just burnt a tonne of trust.

I find it interesting that this reflection didn't mention SLI/SLO/error budgets, which Google SRE has championed for a long time.

My impression is that they're nice in theory, but less useful in practice. I'm yet to see an error budget effectively inform eng decision making.

Other: Node.js, Golang, CDNs, Cloud-managed services, etc.

How is (a) wrong? Pg doesn't horizontally scale like Mongo does

The author also mentions Neon is the first "pioneer" to do disaggregated storage. It is mentioned in the 2006 Bigtable paper, and probably earlier than that too.

It's nice to see another pg proxy using the pgx parser (their src[1]) - I built one using this lib too. However, this implementation is missing a lot of low level features to be considered close to compatible, including: multi-query transactions, auth, TLS, extended query mode, query cancellation.

[1]: https://github.com/zknill/sqledge/blob/main/pkg/pgwire/postg...

SF streets are notably worse than the cities you listed

The original post does describe several sub-problems. The group could first chip away at global state, signals, libraries. They can do this before changing the process model in any way.

To me, the worst Google corpspeak/best meme was "Advancing our amazing bet" https://fiber.google.com/blog/2016/10/advancing-our-amazing-...

In this whole exchange you implied you lived in a 1br apt for $1.6k in a cool area today, but now it comes out you share a 3br. I also chuckled

Deletes (tombstones) shouldn't really get in your way if you don't need them. Similarly, range scans just come for free from SSTables being sorted. Archiving RocksDB SSTable files can be a decent strategy.

One thing to pay attention to is if your telemetry data is indexed by timestamp (i.e. you're writing to the keyspace in order), the compaction of immutable SSTables layers could be wasteful? Although, the author's nice example of non-overlapping SSTables key ranges suggests there may be minimal write amplification here too.

It said the 2nd ranked "Portland" in NYC is Portland, CT (??) when it should almost certainly be Portland, MA

I am pleasantly surprised that Wiz published that screenshot. It looks super confidential.

What is the AIR business unit?

Caches. It is generally a good idea for a data intensive application to use as much RAM as is available for it. Better that than the RAM sitting unused