As the adoption of electric vehicles (EVs) increases, it is important to reliably characterize their recharging load on the electrical grid. Because the last-mile of the electrical grid was never envisioned for EV usage, we need to identify and localize demand-side hot spots due to EV charging. Also, itwould beworthwhile to see howthe EVs can be utilized for the benefit of the grid on activities such as peak reduction and ability to sustain local micro-grids. Quantifying the impacts of EVs on the grid requires an understanding of the spatiotemporal distribution of EVs in a city and the consumption patterns of the EV batteries. These, in turn, depend on the traffic load on the transport grid. In this paper, we attempt to understand these impacts of EV by creating a model of a popular EV (Tesla Model S) and integrating it with SUMO, a broad-based micro traffic simulator. Using this setup, we obtain the EV load on the distribution side of an electrical grid for a real-world traffic pattern dataset from the city of Luxembourg. We find that: (i) The city's aggregate peak demand can be managed within existing levels even when 25% of vehicles become electric. (ii) However, EV charging does overwhelm the distribution network creating hot spots and these hot spots can be clustered together spatially necessitating additional upstream investments. (iii) When EVs feed power back to the grid, simple algorithms can achieve reasonable aggregate peak shaving (~7%) under low EV penetration levels. For higher EV penetration levels, sophisticated EV coordination algorithms are needed. (iv) Under a penetration level of 25%, EVs can potentially sustain micro-grids that serve the entire base load of 13% of the population for a duration of up to 30 minutes.