Another very nice overview video - thank you. And thank you for mentioning the "thundering herd" effect at 1:45. Over the years I've been surprised by how often this issue arises in systems built by people who should know better. And of course this is not just related to Redis.
I think this video missed to talk about the very important fact that Redis has a limited memory space and depending of its configuration, when full, it will erase data. For example, when using as a Session "storage", the default configuration would make it so that in conjunction with TTL, the older/inactive sessions will be destroyed to make place for the new ones, effectively acting like a FIFO. So you have to carefully calculate the average size of your session's data and divide the available Redis space by it to obtain the theorical number of possible concurrent sessions and plan accordingly.
@ByteByteGo thank you for the video. I have a small complaint. Making the graphs bigger over time, makes me dizzy to the point I can't read them (I have to pause the video to read them), maybe this is just me. Everything else is great, love your explanations.
You are one of the best teachers on youtube for this kind of stuff. The animations are also super helpful to understand the content. I think that if you enunciate your words a bit more, itd take your video quality to another level
great video, thanks for sharing! didn't get leaderboard scenario part where you say it helps to retrieve entry by its score ... can't imagine that happening in real life.
In the rate limiter use case, would the rejected request retries certain times until the queue is available or would the server simply return rejected response?
A coworker of mine uses it to compliment Elasticsearch, in cases where certain shards get more hits than others, have not tried it myself but it sounds sensible for dealing with usage spikes.
Amazing work! Don't give up, you'll come through happier and stronger soon. Enjoy your time with the family. Looking forward to March 10th! That's when the course comes out right?
@@pjf7044 I've used sorted liste , each agent has a score depending on how much he waited and his performance after each call you update his score, and whenever you have a request for that call you pick the agent in the top of the list
Redis is always really nice to use but I miss one feature and that is multi-tenancy. It's always kind of a pain (and kind of a waste imo) that I have to use and maintain 1 redis instance per app (or multiple when using replication) instead of setting up 1 cluster, giving each app their own space (with no access to the spaces of other apps) and calling it a day.
Thanks, Sahn for yet again a great video. Out of curiosity though, I am wondering what tools and techniques you use to create these smooth animations for your videos. I have a RU-vid channel of my own and I would like to incorporate these as well. Thanks.
Probably my bad, but why (minute 2:15 and later) you speak about stateless, if you are storing and managing the session id ? It should be stateful ? Thanks (if I am wrong please tell me).
@@gcasar Thanks. That totally makes sense now that u explained. But technically the server + redis as a whole is providing statefull web service. Am i right?