The performance of unipolar unicolor coded Brillouin optical time-domain analysis (BOTDA) is evaluated based on both Simplex and Golay codes. Four major detrimental factors that limit the system performance, including decoded-gain trace distortion, coding pulse power non-uniformity, polarization pulling and higher-order non-local effects, are thoroughly investigated. Through theoretical analysis and an experimental validations, solutions and optimal design conditions for unipolar unicolor coded BOTDA are clearly established. First, a logarithmic normalization approach is proposed to resolve the linear accumulated Brillouin amplification without distortion. Then it is found out that Simplex codes are more robust to pulse power non-uniformity compared to Golay codes; whilst the use of a polarization scrambler must be preferred in comparison to a polarization switch to mitigate uncompensated fading induced by polarization pulling in the decoded traces. These optimal conditions enables the sensing performance only limited by higher-order non-local effects. To secure systematic errors below 1.3 MHz on the Brillouin frequency estimation, while simultaneously reaching the maximum signal-to-noise ratio (SNR), a mathematical model is established to trade-off the key parameters in the design, i.e., the single-pulse Brillouin amplification, code length and probe power. It turns out that the optimal SNR performance depends in inverse proportion on the value of maximum single-pulse Brillouin amplification, which is ultimately determined by the spatial resolution. The analysis here presented is expected to serve as a quantitative guideline to design a distortion-free coded BOTDA system operating at maximum SNR.