I'll save this video for later

We don't have to care about dropping requests, because our app with only

this was the first time I had fun reading and understanding a blog post.

Main thing I am taking away from this video... Sam is a FKKING LEGEND !!!

"... unless it was Jira and it was my job..." 🤣

this article was incredible

We just finished learning priority queues and now working with weighted graphs and trees in school. A lot of people say you don't need to know these things, but they really are used in a lot of applications

The amount of care and polish Sam puts into his posts cannot be understated. Really solid stuff!

absolutely loved this, and thank you for introducing me to Sam! all blogs and web based learning material should be as interactive as this!!

I wrote a queuing system some years ago that allowed a mix of all of those strategies plus some other features. It was for data processing, though. I was converting a large, monolithic batch process (a couple hours to run) to be parallel and closer to realtime (sub second.)

Queueing is Britain's national sport!

Queue the video

I wonder about using multiple queues in parallel? Like the priority queue on 2B2T.

As you mentioned in passing, it would seem trivial to add an initial processing branch that filters "expired/timed-out" requests providing an early exit, and rapid dequeuing of these abandoned requests. Similarly, why isn't there a comparison and drop on enqueue, that prioritizes oldest entries, which at worst would restart the TTL for its queue slot. Yet another thought, dequeue capacity is the metric to gauge - how long does a request take to process? Once that time, multiplied by the depth has elapsed, the system has gone pear shaped and additional capacity should be allocated. I am reminded of all the priority queuing knobs at the network layer (including WRED and RED) that exist but are both largely unused on the "public" net, and used only in limited fashion on corporate networks, to accommodate RTP media traffic. It was telling that in the US and parts of Europe, ISPs strongly preferred simply adding capacity, rather than tinker with queuing. This was in large part because non-FIFO queues were hard to reason about, and turning the knobs only makes a difference on the edge cases - and when I say edge cases, think about the last person being shoved onto the subway by a train attendant in Tokyo. If you wanted a better "experience" then adding capacity, i.e. ability to handle/dequeue requests, is the only game in town. One more thought; Fry's Electronics utilized a single queue for its dozens of checkout counters - effectively a shared scalable processing queue. This was effectively human scale load balancing.

Theo: Check out Sam! Me: Sam who ? Theo: Yeah!

This reminded me a lot of the visualizations on the Nicky Case blog (particularly, the "build a better ballot" one)

I don't think FIFO is from the engineering world though, I think it's originally used as an inventory method in retail for example

What an incredible article!

british person’s dream

FIFO was a term way before engineering. Great vid though!

What about if we have the priority queue but when a priority request arrives we would drop a low priority request from the back of the queue, then add it after the last priority request, that if the queue is not filled with priority requests. I think this may be better then the queues in this article.

great post, great discussion of it

omg the end of the article was so clever

Those interactive examples for teaching stuff - they are GOLDeN :) Good work! Dropped requests - very fun :)

16:40 "Let's be real a priority request should never take 18 seconds. Like imagine hitting a checkout button, and 18 seconds later you finally get to checkout. That's not acceptable" Surprisingly, this is the opposite of how I would feel. While I wouldn't like it, I'm much more forgiving of such delays on stuff like a checkout because at that point I'm committed. I've been trained to not try to refresh the page or hit the button again when doing the scary operations for fear of getting duplicate charges to my cards and what not, so I'm much more likely to actually see the request to completion. But if I'm getting such delays on the non-important stuff, like simply showing me product pages in the first place or on adding stuff to my cart, then I'm much more likely to bounce off the website entirely since I'm not committed to anything yet.

Would be an interesting addition to account for request processing time. E.g. A ping vs something that does a bunch of database stuff / processing.

Your videos are getting high priority in my subscriptions nowadays 😂😂 (dropping other videos)

An engineer became bored, so he became a doctor 🙃

Seems like a priority LIFO would be a smart choice. If the queue is full, you could also drop a normal request inside the queue if a priority request is coming in. Also instead of dropping new requests, maybe you could instead drop the oldest request in the queue, or have a 50/50 chance of dropping either.

How about priority + RED, but LIFO for lowest priority? (Or, we could use LIFO within each priority bracket, especially if we have multiple priority levels.)

I thought you started data structures lol

The thing I'm stuck on is "what is a priority request" at the moment. I guess because I was only thinking of a basic user collab app, there is no priority. I guess if I have moderators, they should have priority at least!

Fun drinking game: take a shot every time he geeks out over the visualizations (no but seriously they are incredible)

I'll watch it now

0:05 speak for yourself, I’m British. 😂

Had to go and search for the other video to make sure I wasn't going crazy... loving Sam Rose's content

Video’s title brings a different vibe in French

doesnt apache kafka handle queues under the hood? // im still learning more about this stuff.

Was FIFO really a tech thing first? I learned about it for the first time prepping food for freebirds (restaurant)

LIFO is not a good option for PUT requests (or any state update), since it makes them out of order. You can't afford to switch cancel and resume actions or updating same field with different data.

your video quality has increased in the last few videos. when you're not constantly shilling vercel and react it's much better.

There's an error with Priority Queue demonstration and explanation. Priority queue will not let a checkout request with high priority drop because it will replace a low priority requests in the queue. That's how bounded PQ works by definition. It's impossible to lose a priority request if there's a low priority request in the queue. You can only lose a priority request if ALL requests in the queue are priority requests.

Really nice video

I have seen this nice interactive one

This post is amazing.

finnaly, a video thumbnail without face

Do people really click off when a page takes more than 5 seconds to load? For me, when doing things like shopping, during checkout the "brain timeout" is way above 30 seconds.

I have one problem with RED: If the queue is below like 5% full or something, instead of dropping 5% of requests, none should be dropped. Why always have a minimum level of failure when you probably would always be able to handle all of those requests?

wait, why are LIFOs better for most workloads? did I miss something? Seems like FIFOs are much better queues.. edit: ok it's because fewer 'timed out' but what scenario would that be the most important consideration in?

damn, this is an insta-save article and video

watch queing video? = 'at least once'

I guess this explains why all the balls at the bottom had stripes, because I kept forcing priority requests to drop just for funsies. Going into the hundreds kind of breaks it, though.

And then there is the elephant in the room: Don't actually process a disconnected or nearly timed-out request. Just drop it instead. This is a trivial optimization and should be the first thing to do. When trying to enque any request while the queue is full, replace the request waiting the longest with the new request. That way, it's still fair FIFO but when not everyone can be served, you still get a chance as the requests in front of you might be about to time out. You totally can scan the entire queue every time a request arrives to determine which one to drop. Worst you get is O(N). It's fine. Don't go all-in optimization fetish on request queues. The CPU shouldn't be the bottleneck in your load balancers.

Drop the timed out requests from the queue, do let them go to the server.

really interesting video, the thumbnail and title don't do it justice. it could benefit from a better title like: "Queues are more interesting than you think", "I bet you didn't know this about queues" and a proper thumbnail visualizing the priority conundrum. What about a thumbnail with the text "don't do this" and a red arrow pointing to the bottom end of the queue

ThE pHp ExPeRiEnCe... Bah! Try sending many(hundreds or thousands or more) requests to a PHP server, and share the percentage of dropped requests.

07:58 what if random request processed?

Caught in time..

If you have a good reach, like Theo, you may cover the work of other specialized devs, like Sam, who have a good content. A mutually benefitial thing :)

For the Americans out there: Lines Thank me later 😂

People lining up to watch this video.

This article grossly misrepresents how long timeout take to process in proportion to requests.

Hi Twitch!

Fantastic visualizations, but the article misses the mark quite a lot IMO. Two tings that do way more than anything in the examples here is: - Drop the the requests that are considered timed out. - Have a separete queue for priority requests and serve that queue first. The visualization is also misrepresenting what would happen in a real life scenario where you have multiple servers/workers processing requests in parallel. The scenarios where the server capacity is so under dimentioned is also very extreme and not realistic at all.

Queues

Open for the listen queue spelling on english 🎉 thank you!

FIFO is pronounced "F-eye Foe" rhyming with High-Low. And rhymes with LIFO.

14:15 bruh really

kwewes

Ok

I am so not the target demographic to tell "waiting longer than 5 seconds makes users leave the page" - I am sorry but I am so used to the internet being spotty at times I will always wait as long as I have to for a page to load. What arrogance leads people to quit interacting with something they actively chose to interact with if they need to wait a bit? Screw statistics and user surveys, if a user quits when they have to wait, I do not want to serve them.

What http servers are so stupid as to serve a request from a client that has already disconnected?

You are so beautiful when you talk about tech ☺☺☺☺☺☺

Why do you say FEFO

Author should receive some prize , together with this Bytes2Go guy