Cloud-based data management platforms often employ multitenant databases, where service providers achieve economies of scale by consolidating multiple tenants on shared servers. In such database systems, a key functionality for service providers is database migration, which is useful for dynamic provisioning, load balancing, and system maintenance. Practical migration solutions have several requirements, including high availability, low performance overhead, and selfmanagement. We present Slacker, an end-to-end database migration system at the middleware level satisfying these requirements. Slacker leverages off-the-shelf hot backup tools to achieve live migration with effectively zero down-time. Additionally, Slacker minimizes the performance impact of migrations on both the migrating tenant and collocated tenants by leveraging ‘migration slack’, or resources that can be used for migration without excessively impacting query latency. We apply a PID controller to this problem, allowing Slacker to automatically detect and exploit migration slack in real time. Using our prototype, we demonstrate that Slacker effectively controls interference during migrations, maintaining latency within 10 % of a given latency target, while still performing migrations rapidly and efficiently.

Data drives knowledge which engenders innovation. Be it personalizing search results, recommending movies or friends, determining which advertisements to display or which coupon to deliver, data is central in improving customer satisfaction and providing a competitive edge. Data, therefore, generates wealth and many modern enterprises are collecting data at the most detailed level possible, resulting in massive and ever-growing data repositories. Such massive scale of data pose a number of research challenges, called big data challenges, which form the basis for my research. My research philosophy is to build database management systems (DBMSs) designed for large scale operations that expose abstractions to simplify application design while providing tools to ease system deployment and management. Using this philosophy as the cornerstone, my research has spanned the broad area of big data management encompassing both transaction processing and analytical processing systems exercising a synergy of both theoretical and practical system-oriented research. 1 Dissertation Research: Scalable, Elastic, and Autonomic OLTP Databases DBMSs serving mission critical user facing web-applications must be scalable, fault-tolerant, and highly available to serve the growing number of users and handle the increasing amounts of data. Classical relational DBMSs (RDBMSs) support generic transactions but are expensive to scale-out to large clusters. Key-value stores can scale-out but provide transactional access to only single key-value pairs, thereby considerably increasing the complexity of application

The success of cloud computing as a platform for deploying webapplications has led to a deluge of applications characterized by small data footprints with unpredictable access patterns. A scalable multitenant database management system (DBMS) is therefore an important component of the software stack for platforms supporting these applications. Elastic load balancing and efficient database migration techniques are key requirements for effective resource utilization and operational cost minimization. Our vision is a DBMS where multitenancy is viewed as virtualization in the database layer, and elasticity is a first class notion with the same stature as scalability, availability etc. We analyze the various models of database multitenancy, formalize the forms of migration, and identify the design space and research goals for an autonomic and elastic multitenant database.

Cloud technology is evolving at a rapid pace with innovation occurring throughout the software stack. While updates to Software-as-a-Service (SaaS) products require a simple push of code to the production servers or platform, updates to the Infrastructure-as-a-Service (IaaS) or Platform-as-a-Service (PaaS) layers require more intricate procedures to prevent disruption to services at higher abstraction layers. In this work we address the need for rolling upgrades to PaaS systems. We do so with the App-Scale PaaS, which is a multi-application, multi-language, multi-infrastructure, and multi-datastore platform. Our design and implementation allows for applications and tenants to be migrated live from one cloud deployment to another with guaranteed transaction semantics and minimal performance degradation. In this paper we motivate the need for PaaS migration support and empirically evaluate migrations between two AppScale deployments using highly scalable datastores. 1