great stuff. Ignore the views - over time feel like this channel will blow up. thanks for the dedication. wish i had the dedication to do something like this not for other people but just for my own knowledge - cant imagine what you have learned from doing all these videos. great job.
No doubt. This is by far the best fast, yet in-depth distributed system design series I've ever come across. It will BLOW!!! I wish there was a way to invest $$$ in this channel :)
Can't agree more! This is some unbelievable learning and sharing! Very inspiring! Jordan, is that the reason why you "have no life" cuz you spent all your time on this and working out? haha
I wish you would do more videos explaining things like this. I'm a Techpreneur and I'm trying to create a ride-sharing business and there are lots of details that this video has put in context for me. Thank you very much Jordan.
One of the most interesting aspects of working with the mapping groups at some of those companies was their pathological stubborn refusal to consider using off the shelf/best-in-class solutions, like Postgres with PostGIS extensions, along with using GDAL/OGR as their geospatial operations library. It was like watching someone pour gasoline on a huge pile of cash and lighting it on fire. Every month. Some even refused to geo-align their collected data with a widely used datum, EPSG:4326(WGS-84 w/recent update) so they couldn’t import data from other sources OR generate revenue by selling geodata to potential customers. What a MCF.
Super dope content! You should create content on Leetcode patterns and cracking coding interviews after System Design, and then this channel would most likely become the premium one stop shop for coders world over for interview prep!
Yeah, i may do this. I don't love the idea of making redundant videos, as there are plenty of them on LeetCode, but ultimately I may just start with Leetcode and then get into more complicated topics. I've wanted to get into competitive programming anyways, so perhaps making videos on it would help me learn.
Great Content! It was super helpful. One small suggestions would be In your title when viewing on playlist, we are not able to half of the title. if you can start with topic and then mention google SWE. I think it will be helpful. Thanks
Did not follow the range query across geo hashes, in the slide where we were calculating the nearest points (slide having the diag of boxes)? Geo hashes don't have a mutual ordering - or do they? (if you don't mind, can you share a bit on that)
You're right that there's not a direct ordering, but for a given geohash I do know all of the 8 hashes of the surrounding boxes so I can just check all of those with a Pythagorean equation to check things are in the correct distance from my focal point
Well yes and no haha - from a systems design perspective I'll probably do it as a practice question eventually. From a machine learning perspective, probably not just because that's a whole different area than what I'm covering.
If you have a multi column index why would it be a full table scan to ensure that the longitude is within the correct bounds also just as the latitude must be?
So if my business is only serving a particular region, say Nigeria or Africa, then it wouldn't make sense storing ALL the geohashes of all coordinates in the globe right? It would be better only persist a subset to Redis. Counter to that, it seems this that if we have a distributed Redis cache totaling 1 TB for instance, it would not hurt to save everything. Penny wise, Pound foolish :)
great stuff. Ignore the views - over time feel like this channel will blow up. thanks for the dedication. wish i had the dedication to do something like this not for other people but just for my own knowledge - cant imagine what you have learned from doing all these videos. great job.
Thanks Eric, means a lot!! I'm definitely learning a ton and having fun in the progress, hope you're getting something out of it as well!
No doubt. This is by far the best fast, yet in-depth distributed system design series I've ever come across. It will BLOW!!!
I wish there was a way to invest $$$ in this channel :)
Can't agree more! This is some unbelievable learning and sharing! Very inspiring! Jordan, is that the reason why you "have no life" cuz you spent all your time on this and working out? haha
yep
Best material on UA-cam on the distributed systems. Thanks!
I appreciate it
I wish you would do more videos explaining things like this. I'm a Techpreneur and I'm trying to create a ride-sharing business and there are lots of details that this video has put in context for me. Thank you very much Jordan.
The content and style is really good. Thanks for the effort and keep up the good work!
Thanks for watching! The positive feedback keeps me motivated haha
One of the most interesting aspects of working with the mapping groups at some of those companies was their pathological stubborn refusal to consider using off the shelf/best-in-class solutions, like Postgres with PostGIS extensions, along with using GDAL/OGR as their geospatial operations library.
It was like watching someone pour gasoline on a huge pile of cash and lighting it on fire. Every month.
Some even refused to geo-align their collected data with a widely used datum, EPSG:4326(WGS-84 w/recent update) so they couldn’t import data from other sources OR generate revenue by selling geodata to potential customers. What a MCF.
Haha, that sounds painful!
Super dope content! You should create content on Leetcode patterns and cracking coding interviews after System Design, and then this channel would most likely become the premium one stop shop for coders world over for interview prep!
Yeah, i may do this. I don't love the idea of making redundant videos, as there are plenty of them on LeetCode, but ultimately I may just start with Leetcode and then get into more complicated topics. I've wanted to get into competitive programming anyways, so perhaps making videos on it would help me learn.
Thank you
Great Content! It was super helpful. One small suggestions would be In your title when viewing on playlist, we are not able to half of the title. if you can start with topic and then mention google SWE. I think it will be helpful. Thanks
Yep I've been doing this recently
Very nicely explained, great explaination!
Did not follow the range query across geo hashes, in the slide where we were calculating the nearest points (slide having the diag of boxes)? Geo hashes don't have a mutual ordering - or do they? (if you don't mind, can you share a bit on that)
You're right that there's not a direct ordering, but for a given geohash I do know all of the 8 hashes of the surrounding boxes so I can just check all of those with a Pythagorean equation to check things are in the correct distance from my focal point
Great stuff explained in very easy manner
can you make a video on recommendation systems?
Well yes and no haha - from a systems design perspective I'll probably do it as a practice question eventually. From a machine learning perspective, probably not just because that's a whole different area than what I'm covering.
I am lucky to get this video.
awesome content ...great work
Thanks!
thank you!
great video!!
If you have a multi column index why would it be a full table scan to ensure that the longitude is within the correct bounds also just as the latitude must be?
It's linear in the worst case when all points are within the latitude that you care about but spread out all over the place with longitude
How can we cache hot searches? I think we can have another Redis instance within same node that can cache hot searches (Geohashes) using LRU.
That's reasonable, for sure!
want to thumb up but really dont wanna break that 256 likes..
So if my business is only serving a particular region, say Nigeria or Africa, then it wouldn't make sense storing ALL the geohashes of all coordinates in the globe right? It would be better only persist a subset to Redis. Counter to that, it seems this that if we have a distributed Redis cache totaling 1 TB for instance, it would not hurt to save everything. Penny wise, Pound foolish :)
Yeah I'd have to say that if you only need to index a certain area then only store hashes for that area.