mmontagna9

If you're interested in using them you should just bench them yourself.

Aren't we all...

People run postgres at scale which is much "larger" than what you are running, which isn't to say that your workload isn't substantial, it is. But there are folks who push it harder, both in terms of writes/rows and workload complexity. It's one thing to write a ton of rows into a single table, it's another thing entirely to write into many tables, index all of them and then query them at scale; you didn't mention much about your workload complexity so I'm not trying to suggest it isn't complex, but there are certainly plenty of folks on this forum who can attest to it being tricky to scale high write workloads, in large organizations with many stakeholders and complex query loads.

The youngest old man I know at any rate

Managing one pair of headphones that connects to more than one device is a nightmare, everything always auto connects/disconnects when you don't want. Many devices don't have a temporarily disable connection attempts toggle.

Sadly this isn't true. Postgres will happily replicate and skip data if you tell it too.

And there have been multiple bugs around logical replication in version ~10-15 that can cause data loss. None of these are directly related to lsn fiddling tho.

You caught our off by one bug :)

Where are you getting your numbers?!

Yeah there isn't much publically available. We wrote a tool for this at Instacart, it relies on some internal tooling but much of it is generalizable (assuming you run pgbouncers). This is a good reminder for us to write a blog post or something.

Needs a clearly marked manual kill switch

> Reduced maintenance needs. OrioleDB implements the concepts of undo log and page-mergins, eliminating the need for dedicated garbage collection processes. Additionally, OrioleDB implements default 64-bit transaction identifiers, thus eliminating the well-known and painful wraparound problem.

I love this, no more vacuums?

> In my experience, PG is always the most out-of-date piece of any system.

Yes

Some shops don't design like that :) some folks build giant monoliths and then have to scale/deploy that app with it's database connections broadly.

Yep the readme is pretty clear it's a POC "Alpha: don't use in production just yet."

That said I've struggled with more than my fair share of pgbouncer misconfigurations and outright bugs. I personally would love a simpler (harder to misconfigure) and less buggy alternative.

This is great. Let me know how I can help

I don't think you need a cleanroom if your feature sizes are huge.

Tell that to my 70 year old grandmother who has replaced both the motherboard and hard drive in her Lenovo desktop. If you design a somewhat reasonably easy to repair computer people are more than happy to fix it themselves. It's only because of the difficulty of repairing apple products that people shy away from the idea.

We've been tracking this SDK at MightySignal. I can confirm that the apps often record data for some time after the app becomes inactive and that the amount of data sent is enough to represent a full audio stream, multiple requests with sizable blobs per second, bothers me mostly because of the terrible drain this SDK must be putting on user's batteries. It's only present in ~280 low quality android games atm.

Interesting, looks like a fair number of apps use webrtc directly.

https://mightysignal.com/top-android-sdks?tag=47 https://mightysignal.com/top-ios-sdks?tag=47

Yes we are getting slowdowns and internal errors as well. Also noticed one of our EC2 instances shutdown due to failed health checks.

We've experienced agent crashes in the past but those seem to have been resolved now. Occasionally we will find a docker container from an old task which is still running, but about which ECS knows nothing. Definitely can make for an interesting troubleshooting adventure.

And it seems like maybe the ECS team is trying to move a little too fast recently. They released this blog which claims the run-task api supports several new override parameters but the backend still doesn't actually do anything with them it just silently ignores them.

https://aws.amazon.com/about-aws/whats-new/2017/06/amazon-ec... https://github.com/boto/boto3/issues/1184

"...Grace Murray Hopper was a U.S. Naval officer, and an early computer programmer. According to him..." Grace Hopper was not a guy! Don't call her a him!

The only requirement is the servers are actually located in the respective countries. Everything I've found so far either requires a colo or the datacenter is actually in the US or Brazil.