As outlined for instance by the CAP theorem, achieving consistency guarantees within a 100% available and faulttolerant distributed system is impossible. Nevertheless, in real-life applications actual properties are neither black nor white and the degree of fulfilment of requirements depends on the likelihood of failures and communication parameters of distributed systems. While typical Cloud-based applications weaken consistency in accordance with less strict applications requirements, strong consistency can also be achieved, for instance by tunable consistency. This, however, often comes with a strong degradation of scalability (performance of growing clusters) and availability. Based on a project investigating the usefulness of Cloud DBMS for Massively Multi-player Online Role-Playing Games (MMORPGs) we describe how strong consistency can be provided for such a scenario, by still proving a high-level of availability and performance suitable for this specific application. For this purpose we implement a lightweight mechanism to detect failures based on timestamps and only react accordingly if required.