tweenagedream

Which ones?

According to https://support.google.com/pixelphone/answer/4457705?hl=en pixel 5a is getting eol this month, with the next security update dropping for pixel 6 starting in October 2026

"Last gen" pixel 8 is going to get android and security updates through October 2030

Who says you have to purchase a service with dollars?

I mean they did say almost no breaking changes, depending how much has changed in the language, a small handful of breaking changes in niche use cases may be considered almost none by some. I'm not sure I would say the comment is absolutely inaccurate.

I experienced this just last month! Except, I didn't know it at the moment. I was participating in a design review while wfh on meet. We were having some good discussions, which I participated in. Partway through the meeting I start to fidget a bit and unplug my mic, untwist it from my headphone cable and plug it back in. Next time I unmute, unbeknownst to me, everyone else starts hearing the local radio station. I was using a schiit fulla DAC, and one of its features is auto gain on the microphone. Well I apparently didn't plug the mic in enough and it picked up the local radio station and then boosted the gain enough that it came through loud and clear over the call. Blew my mind once I learned wtf was going on. Embarrassing in the moment as I was muted twice and told to stop listening to a podcast during the meeting.

Can you please link to the incident where the "full region" was down for weeks?

This is because of the way tcp congestion control works. In order to achieve higher speeds, tcp has a "window" that grows, essentially the amount of data in flight, that has not yet been acknowledged by your client. That window doesn't open very wide in the 100kb test because it happens so quickly.

I don't think that is true. MKBHD just came out with a video demonstrating using a video wall that updated in sync with the camera frame rate. Allowing two cameras to see different backgrounds, e.g. two different colors or even two different images, as the demo was showing a parallax effect.

For gRPC, the protobuf is the API specification, a service definition with endpoints, what requests to those endpoints should look like and what responses look like. Of course, there are better and worse implementations, e.g. a well commented proto definition explaining what various args do, etc.

In gRPC the definition is a requirement to use, so at the bare minimum you have the typed structure of requests and responses. There is no such requirement for REST

I work at Google so am perhaps biased, but that article is from 2019 and cloud is now a 20B/yr business and growing. Why does it keep coming up?

Usually a connection is defined by what's called a 5-tuple. Source up, source port, dest up, dest port, protocol (e.g. tcp or udp)

Indeed, it's a publicly documented feature: https://support.google.com/mail/answer/7436150?hl=en#:~:text....

I think down leveling is quite common, especially at higher levels as the expectations grow quite a bit between senior and staff. Depending on the position, actually filling it with +/-1 of the target level is a win. Interviewing is expensive and if a candidate is good for the role, but perhaps just a bit below the bar, then a recommendation to hire at L-1 tries to keep the candidate and give them a career path.

This is from my perspective at Google, I've never worked at Amazon.

Seems like there are quite a few options that show up pretty quickly when searching for multipath udp e.g: https://github.com/angt/glorytun

Would that work for you?

I was a simple customer since before the BBVA acquisition, but once I heard that they were shutting down the main service I moved everything over to a cash reserve account in betterment, which I already had for investing. Since then I've opened an m1 finance account which had a 1% interest rate and 1% cash back if you use their debit card. (I tend to only use credit cards since the rewards are better)

Closed my account a few weeks ago. sad to see simple go, it was refreshingly easy and just worked, had great features and was something I never thought about since it just worked.

Generally RAM doesn't burn out, so I don't think that will be the reason.

My guess it will be around capacity, though if the LEO ISP market has multiple players all covering the same areas, there is a chance we don't fall into the market segmentation we have now since there will be multiple options.

DNS doesn't really redirect, but DNS is the answer. What you're talking about is split horizon, this is most common when you have an authoritative server for say a corporation, but can be applied here as well.

That said, I have a local dns server that is pi hole, I've set aliases in that that are preferred over my domains dns which would be used outside my network.

That way you can have pi hole pint to 192.168.0.0/24 and Google domains or whomever point to your vps.

Except, sometimes a data center goes down and no one notices. So if a data center goes down, and no user around notices, is it really down?

Yes.

For the most part, other services take care of network resources. Borg might have some host level network limits, but I haven't seen them used much.

I've asked that my number be unable to be changed to a different sim over the phone. So to change it I need to go into a physical store and present proper ID

Usually grpc connections are left open after a message is sent and received. And since it's based on h2, multiple message streams can be multiplexed over a single connection. That is part of the benefit of grpc, establish a long lived connection and send your messages as needed. Similar to using keep-alive with http.

I've reached out to some folks on the gtm team to look into this.

You have the gist I would say. It's important to understand that Google separates the control plane and data plane, so if you think of the internet, routing tables and bgp are the control part and the hardware, switching, and links are data plane. Often times those two are combined in one device. At Google, they are not.

So the part that sets up the routing tables talking to some global network service went down.

They talk about some of the network topology in this paper: https://ai.google/research/pubs/pub43837

It might be a little dated but it should help with some of the concepts.

Disclosure: I work at Google

What does your stack look like?

It's hard to tailor a postmortem like this to everyone's individual experience but it is surprising to me that your experience is so different.

Isolation is often done in software, Google has invested a lot of effort in making sure the distinct services that they run e.g. Youtube transcoding on the same machine search is running on don't interfere with each other. Whether through cpu constrains or some other priority levels. These are features of borg.

https://scholar.google.com/scholar?lr&ie=UTF-8&oe=UTF-8&q=La...

This is a problem in the web ui creating additional node pools, there is a very simple workaround to use gcloud. What major impact are you referring to?

It's likely the fix is checked in and will start roll out on Monday.

Disclaimer: I work on Google Cloud and while I believe we could use more words here, this doesn't seem like a huge problem. It's embarrassing the the issue with the ui was shipped, and I'm sure this will be addressed in the post mortem as well as whether it could have been mitigated quicker than a roll forward.

This seems to be a test that may turn into a future cloud product? Google has a lot of Pops which provide great latency for users, If they start filling these pops with gpus so that companies like ubisoft or epic or $gamecompany can pay for the infrastructure and let users pay them a subscription and then I don't have to drop 500+ dollars every year to stay relevant with AAA titles, maybe it all works out?

I bet there is already a regression test for this one.

Disclaimer: I work on Google Cloud so I will be speaking from the bias of knowing those products.

They talk a lot about reducing operating complexity and scaling their infrastructure, I wonder what the cost of their current infrastructure + the staff to maintain it might be vs the managed solutions that cloud providers offer now.

For example, using cloud datastore or spanner or big table as a persistent layer, these managed services can definitely scale to the current need and I've seen them go much higher as well.

For logs ingestion and analysis, big query can be a very powerful tool as well, and with streaming inserts that data can be queried in near real time. For things that are less urgent, batch queries. For other things dataflow can help with streaming workloads.

I think one of the problems they alluded to though was that at the moment they're on a single provider, and what they're looking for is a multi cloud strategy which totally makes sense. A lot of the above products create some kind of locking, with some exceptions, like using hbase as an interface to big table or beam as an interface to dataflow. Though I don't know what the other providers offer that may have these same interfaces.

Another option is kubernetes, which I believe all providers are pretty strongly embracing. Having most of the supporting infrastructure be brought up with a few kubectl commands could help them scale across several cloud providers quickly.

Slight correction, they let an ml algorithm control the cooling system in the data center, not Borg. Borg is what schedules Google's software tasks, this was not changed during this experiment.