Using doubly linked list is definitely required. During the get operations you’ll need to move the element to the head and delete from current location. With doubly linked list, deletion will constant time operation. With linkedlist, this will O(n) operation.
I would prefer a query router that would sit in bw Service Hosts and Cache cluster to route the queries, and service hosts and connect to this Router, just like we have in DB Sharding. Would this be a better approach to be suggested in an interview?
I didn’t understand completely that why read requests would go to read replicas. We can read from primary and data will be more consistent. What’s the issue in this approach?
23:35 I would prefer a query router of some sort that would sit bw Service Hosts and Cache cluster and Service hostts just need to connect to the router, the way we have in DB sharding. Would this be a better approach to be suggested in an interview?
Make sure you're interview-ready with Exponent's system design interview prep course: bit.ly/474ucRM
Using doubly linked list is definitely required. During the get operations you’ll need to move the element to the head and delete from current location. With doubly linked list, deletion will constant time operation. With linkedlist, this will O(n) operation.
Please add chapters to your videos. Will help a lot!
Great Video
Thanks, great content! Out of curiosity, what software and hardware (is it a stylus?) is Ravi using to conduct the interview?
It is a digital pen - Wacom pen tablet and software is whimsical
Hey SpyrosGiann, thanks for the kind words! The online whiteboard tool being used is called "Whimsical".
I would prefer a query router that would sit in bw Service Hosts and Cache cluster to route the queries, and service hosts and connect to this Router, just like we have in DB Sharding.
Would this be a better approach to be suggested in an interview?
But why consitency and availability can’t be provided sql databases does that. Cap says any two at one point of time we can support.
I didn’t understand completely that why read requests would go to read replicas. We can read from primary and data will be more consistent. What’s the issue in this approach?
23:35 I would prefer a query router of some sort that would sit bw Service Hosts and Cache cluster and Service hostts just need to connect to the router, the way we have in DB sharding.
Would this be a better approach to be suggested in an interview?
Great content. Love it :)
watched.😀-------