Create the Perfect Uber Bill Format for Facilities with airSlate SignNow

Uber bill format for Facilities

architecture of a ride hailing service requirements functional request a trip find drivers in proximity location updates technical real-time updates efficient driver Rider matching scaling here are some of the messages that will be exchanged between the driver writer and the backend server driver sends location updates every few seconds accepts reject trips offered to him send script status updates once a ride starts Rider sends location updates when he comes online and sends trip request server sends trip offer trip details to the driver and sends information on nearby drivers and trip details to the rider these are some examples of messages exchanged between various entities at a high level we need to track driver location Rider location trip progress match driver to Rider calculate ETA do billing and so on here are some of the services that we would need a location service to track and search locations a trip tracker service to track trip progress A dispatch service to find nearby driver for a rider and ETA service to calculate ETA between two locations a billing service to do the billing or notification service to send messages to drivers and Riders each of these services will be implemented as a micro service let's look at some interactions between the actors and various Services driver sends location updates to the location service every few seconds so the location service is aware about the coordinates of every driver a Riders and strip request to the dispatch service which in turn calls the location service to find drivers that are in proximity of the rider once it has that information trip is offered to a nearby driver when accepted trip begins ETA service may be invoked to find expected time needed to complete the trip trip status updates are sent to trip tracker service every few seconds when the trip ends billing service gets the trip details like route Etc and calculates the amount to be charged this information is sent to the Rider and the driver via the notification service this of course is a simplified flow to give you an idea of the message exchanges each of these Services need to store data so let's add databases here the location service uses a location cache for example redis trip trackers would store information in a nosql database like Cassandra while driver and Rider profile details could reside in a rdbms like MySQL you could use any other appropriate database of your choice here we will look at these services and databases in more detail soon so here's the complete architecture a few additional services like websockets Gateway and a messaging service behind it have been added client to server communication would happen over web sockets which is a two-way real-time asynchronous communication protocol messaging server provides a layer of abstraction between the websockets Gateway and micro Services notification service also uses websockets gateway to send asynchronous messages over websockets to client applications in addition we have added a few analytics related services to the architecture same architecture using AWS Services behind API Gateway is sqs messaging service microservices are implemented as Lambda functions these could also be implemented as containerized applications let's look at some important design aspects now location tracking what does location tracking do it tracks location of drivers and Riders finds drivers in proximity of the rider efficiently but how does it do it here's a map of a city since driver app sends Us location coordinates every few seconds we know where each driver is on this map every driver is uniquely identified in a system with a driver ID now if we want to search for drivers closest to a rider we would have to scan the entire geographical area and examine every single driver that is clearly not efficient however if we divide this geographical area into rectangles or hexagons let's call them cells we would know which drivers are in which cells for example cell 1 has drivers D1 and E2 cell 2 has driver D3 and so on therefore it becomes easier to find drivers close to the rider in this case Rider is in cell 8 and there is a driver D15 in the same cell so driver D15 is a potential match for a trip for this Rider however we may want to cover adjacent cells too for various reasons like a driver in an adjacent cell might actually be closer to the rider and so on so now we have a list of cells and drivers in them to select from and we can use various criteria to prioritize drivers for the trip so this is how we narrow down our driver search efficiently but how do we implement this technically we could use a library like geohash or S2 geometry for this purpose these libraries divide the entire Earth's surface into various cells and every cell is given a unique ID cell size can be predetermined by us given a latitude longitude geohash Library can map it as belonging to a particular cell what's the primary difference between geohash and S2 Library geohash treats the surface it is mapping into cells as flat while S2 can factor in 3D nature of the surface for example spherical Earth U hash or S2 is just a library we still need to store the location information somewhere essentially store list of cells and drivers in them in a real-time fashion and in memory cache is ideal for storing this information for fast storage and retrieval for example a redis cache it stores data in key value pairs if we think in terms of key value pairs we could have cell ID as the key and list of drivers along with car type and free seats in them as the value additionally we could store driver as a key and have cell ID as value having both cell ID and Driver as separate Keys allows us to look up information by either one of them as long as a driver is within a cell everything is straightforward but what happens when a driver moves from one cell to another how do we find out his previous cell so that he can be removed from that there are couple of ways to handle this one way is the driver always sends previous and current location another way is we can look up driver's current cell first before proceeding with updates there may be other ways to store and manage this information however this should give you a good idea of how to go about it we cannot store all the cell driver location data on a single cache server radius in our case we must have a way of scaling horizontally by Distributing data across many servers if you treat a group of cells as a region you can have cash servers by region this allows you to distribute your location data across cache servers by region and therefore scale horizontally to understand this better let's look at the driver location update flow driver sends location information that is latitude and longitude to the server the server converts latitude and longitude to cell ID using jio hash and then looks up region ID based on Cell ID and finally updates region cache server now let's look at write a trip request flow write a send script request from a location that is latitude and longitude server converts the location to a cell ID using jio hash and then looks up region ID based on Cell ID and finally it queries region cache server for drivers in those cells So This Is How We Do driver to Rider matching efficiently database stores trip status updates this could be a nosql database like Cassandra here's what the trips table in Cassandra may look like if you look at rows with trip ID t77 you will notice that updates are received every 5 Seconds driver D15 is driving and Rider r91 is the passenger location or status updates may change with every update you could structure this information differently for example as a Json document if you are using a database that supports that assuming Cassandra database is used in order to scale you will have a sharded setup it's important to choose the partition key carefully for example it could be trip ID how do you calculate ETA between two locations you could use Google Maps distance metrics API it takes two and from locations as main inputs and returns time needed to travel between them this can factor in traffic conditions and so on alternately you could use any other similar service so that brings us to the end of our discussion this should give you a good idea of how to architect a ride hailing service