Approach to build multi-tenant SaaS application with monitoring and SLA

Abstract

SaaS (Software as a Service) is a modern approach to deliver large scalable enterprise software as a service on Internet. Cloud computing platform provides the scalability, availability and utility computing for services on internet. There are many technical challenges involved in SaaS development. One of them is multi-tenancy, which allows single instance of software to serve multiple organizations by accommodating their unique requirements through conguration at the same time. SaaS architecture requires both conguration and some level of customization to achieve higher maturity model. In this thesis, we propose a metadata based SaaS application architecture which is independent of underlying cloud infrastructure. We aim to propose independent SaaS platform concepts, to avoid vendor locking as observed in case of many commercial service providers. SaaS application development should be independent of underlying infrastructure so that application can be migrated from one cloud to another cloud without changing the code. It is possible only if all the players follow the identical as well dened standard SaaS architecture. Our proposed architecture includes monitoring, tenant management, tenant administration, tenant conguration and large data management services. Existing architecture has used simple XML le to store and retrieve tenant specic conguration. File operations are the bottle neck while accessing data for large organization at the same time. In this approach, we have used Memcached concept and it is supported by almost all databases to boost the performance. In addition to this, we have considered application pooling on a web server to manage priority among tenants. Application pooling works as a static load balancer for incoming large request. To realize proposed architecture, we have developed and demonstrated seleced functionalities of University Management System and it is capable to support multi-tenancy.