Difficult when the options are Medium and Large!

This also requires knowing how much it will shrink, and accurately gauging if I've left enough buffer when trying something on at the store.

What would you consider to be small or medium? I have a use case for analytics on ~1 billion rows that are about 1TB in postgres. Have you tried on that volume?

We haven't tested this with 1TB Postgres databases yet, assuming that most companies operating at this scale already built analytics data pipelines :) I'm curious if you currently move the data from this Postgres to somewhere else, or not yet?

Not yet, mostly just kicked the can down the road due to costs. Like you said in another post, careful indexes on postgres get you quite far, but not nearly as flexible as a columnar DB.

I think your project is great. I suspect incremental updates will be a big feature for most uptake (one we would need to try this out at least).

Agree! What got me into programming was building/running my own Counter-Strike server and corresponding stats website. Now with companies running their own servers internally, that isn't nearly as common.

I thought so too, but then was surprised to see:

   Based on the activity logs provided by Okta for their Support Portal, the HAR file had not been accessed by their support engineer until after the events of the incident

That tells me either the attacker either removed the access from the logs OR the attacker figured out how to access the HAR file without having the access recorded in the logs.

All this says is that the support engineer didn't access the file. It doesn't say the attacker wasn't logged accessing it.

>Oct 21, 2023: Okta confirmed publicly that their internal support systems were compromised. This answers how the HAR file was accessed by the attacker and that the initial compromise was not through the employee’s laptop.

how can "the industry" trust okta?!

The social app uses .net - Threads.net

This is where I'm confused as well and yours is the first comment mentioning it.

A more plausible scenario is that IT reset their AD/IDP password because of the lost device and gave them the temp pw over slack + some MFA token.

It's probably because the subset of people who would fall for it likely don't have BTC.

>> In our use case, having attribute model as the primary key in one of the GSIs (Global Secondary Index) which always indicated the type of row was very helpful. With a simple query where model was a hashKey we could get all Members, Channels, Roles, Audiences, etc.

Won't this cause problems with having every item of a type on a single node in AWS since the hash keys are the same? Or, are they suggesting that on the GSI they use KEYS_ONLY, and even though every item lives on one node, the size of all are only the keys (even so, I don't see how that's very useful short of counting number of items).

I think this doesn’t apply to Global Secondary Indexes, but I might be wrong.

Yeah, partition size is only limited if you create a local index, because those are strongly consistent.

Ah interesting, it looks like you are right. If the any partition grows to > 10GB on the main table or a GSI, that partition splits into sub-partitions using the sort key as part of the hashing function. If there is no sort key, the partitioning scheme equally distributes items across partitions, so all partitions sub-divide at the same time.

https://stackoverflow.com/questions/40272600/is-there-a-dyna...

Couldn’t this still hit partition throughput limits though?

It's pretty opaque. My impression is that each shard (storage backend quorum) gets a roughly equal share of capacity you pay for, and items in the same partition tend to live on the same shard to keep range queries small (and local indices require one-shard partitions). They've made improvements in loaning cold shards' unused capacity to hot shards, but they still recommend avoiding hot partitions and keeping load roughly even.

Just as an FYI, when you use code formatting for text it makes the text unreadable on mobile devices.

Not much better on desktop if you leave out line breaks (as was done here).

Thanks, fixed it I think. I tried to put back the line breaks, but it disconnected it from the original text.

Doubt it, the website says it was built 'from the ground up' for a VR experience.

Emphatically not. The interaction space of VR is entirely different from the interaction space of a standard mouse-and-keyboard FPS. For starters, players in a VR game have two hands. A lot of games in VR could have been FPSs instead, but only because they're generally vastly under-utilizing the dimension space that two VR input controllers (plus the input of look direction) give a designer.

If you used the dimensionality of the VR space and you port it to the FPS space, you'll leave entire game mechanics on the floor.

That makes sense. You could use the scenery from the game (although you would probably provide more for more open world gameplay than in the VR game) and also create FPS specific mechanis such as open world puzzles etc.

It doesn't want a squeeze though, it needs resistance/feedback on turning the wheel. Having your hands on the wheel isn't enough, you actually have to give it a little tug, so the alert happens pretty often unless you have a good cadence of tugging on the wheel randomly while in Autopilot.

Hence the increasingly common tactic of Tesla owners to hang a water bottle from one side of the wheel, which provides constant torque, and fools the system into thinking you are always holding it.

That sounds almost dangerous.

Might those tugs affect its learning?

It is NOT learning.