Towards a scalable and unified REST API for cloud data stores

Abstract

In the last years, many database-as-a-service (DBaaS) systems have started to offer their functionalities through REST APIs. Examples are record stores like DynamoDB and Azure Tables, object stores such as Amazon S3 as well as many NoSQL database systems, for instance Riak, CouchDB and ElasticSearch. Yet today, there has been no systematic effort on deriving a unified REST interface which takes into account the different data models, schemas, consistency concepts, transactions, access-control mechanisms and query languages to expose cloud data stores through a common interface without restricting their functionality. This work motivates the design of such a REST API as well as the challenges of providing it in an extensible, scalable and highly-available fashion. To this end, we propose the REST middleware ORESTES that consists of an independently scalable tier of HTTP servers that map the unified REST API to aggregate-oriented (NoSQL) data stores. It extracts a wide range of DBaaS concerns (e.g. schema management and access control) and provides them in a modular, database-independent fashion at the middleware-level. To tackle the latency problem of cloud-based web applications we introduce the Bloom filter-bounded staleness cache consistency algorithm. It leverages the global web caching infrastructure for geo-replication to allow consistent low latency reads. We furthermore show the first steps towards a Polyglot Persistence Mediator that exploits the decoupling of the REST API from the data store to route data and operations based on SLAs.