There is a growing interest in monitoring wide-area power grids, where dedicated devices, so-called phasor measurement units (PMUs), located in substations within the grid, steadily provide measurement data to control centers, thereby enabling the continuous supervision of the system state. The trend towards increasing numbers of PMUs to provide a more accurate and finer-grained system view gives rise to scalability and availability challenges that current monitoring systems cannot handle. Strict latency requirements for the delivery of measurement data further exacerbates the measurement data acquisition and processing problem. We propose a new distributed system using an integrated cloud-based approach for PMU data acquisition, processing, and storage, capable of scaling to large numbers of PMUs while exhibiting robustness in the face of failures. Unlike other wide-area monitoring systems, our proposed system runs on a distributed stream processing platform, thereby making it eligible for cloud-based environments. Our evaluation, which focuses on large-scale data acquisition, shows that the system is able to meet stringent latency requirements and bounds on data loss at a large scale while scaling out the underlying processing infrastructure.