This is an application platform as a service (aPaas) platform whereby an app developer can build and deploy app as a set of independent services that run in response to events and auto-scale for app user. This is being build with the same concept as with AWS Lambda. It supports Function as a service. If user can break his/her application into multiple independent stateless functions, then this platform provides end-to-end support for hosting and running those functions. Platform provides a simple interface to submit the source code files and it will in return provide the user the end-point where he can connect to access that service/function. For ease of development of services which can be run on this platform, platform provides many readymade services which can easily be invoked just like a function call. These are called Class-A services. These services make the application development really fast and easy where the user can easily code a mere function and instantly test it. Services provided by the platform: 1. Authentication/Authrorization and light-session management via token is inbuilt provided by the platform. 2. Automatic load balancing (spawning more processes of a service) is also provided and is based on the incoming traffic for a service. Load is distributed on a number of machines, each being part of the whole system. a) Load Balancing can happen at Hardware level (provisioning more machines at runtime) and at the Service level too (spawning more instances of the same service). b) This functionality makes it a highly scalable. If load increases machine and service instances are provisioned and as soon as it drops, instances are relieved to make it cost efficient. 3. Logging service provides the app-developer a seamless interface to log all the events occuring to his service. 4. SendMail service is an easy to use service to send an emal. Technologies Used: 1. Java 2. Vagrant for automatic machine provisioning 3. Docker as container for a function/service 4. RabbitMQ as messaging queue 5. NFS for file sharing among machines 6. Apache Tomcat for hosting API gateway (end point listener) 7. LDAP for Authorization and Authentication Organization of the platform: Platform extensively uses Messaging Queues for inter-component communication. We have used RabbitMQ. User functions or services are ran inside docker containers which make them feel the isolation and simulatenously provides the much needed security. Various components of the platform: 1. API gateway - the gateway which listens to all the service requests (listens to all service end-points) 2. Load Balancer - Periodically checks and manages the loads on machines as well as services. Flexibly scales the service and machine instances 3. Service Life Cycle Manager - Manages the whole lifecycle of a service - from initialization to call and then cleanup after its completion 4. Server Life Cycle Manager - Manages the while cycle of a server, provisions a new machine if needs and shuts one down if not needed anymore 5. Monitoring Manager - Monitors the machines and MQs and gather stats to be used in decision making of the load balancer 6. Deployment Manager - Handles the deployment of user (App-developer) services onto the platform Why is it called serverless? Platform itself scales up if needed. There aren't any specified servers where the platform might always be running all the time - it varies according to the traffic. It creates an illusion as if the service is not at all running on a server - and hence the name serverless. Contributors: https://github.com/kanishtha https://github.com/Btanwani77 https://github.com/vinigarg https://github.com/Dhawnit https://github.com/hemantverma1 https://github.com/ishan0445 https://github.com/siddharthajain https://github.com/riddhi2000 https://github.com/VikasRaghuwanshi https://github.com/Shikharani29
rids15/ServerlessPlatform
An application-Platform-as-a-service (aPaas) which supports Function-as-a-service (Faas)
JavaApache-2.0