Satellite-Aerial-Ground Integrated Networks (SA-GINs) and Low Earth Orbit (LEO) satellites are emerging as an essential network architecture for next generation mobile networking, such as 6G and beyond. LEO satellites and SAGINs can flexibly configure different network components in tandem depending on the weather conditions, using High Altitude Platform Station (HAPS) and Reconfigurable Intelligent Surface (RIS) assisted Unmanned Aerial Vehicle (UAV). Network configurability is furthermore enabled by deploying both the Free-Space Optical (FSO) and Radio Frequency (RF) links. In SAGIN, since signal-to-noise ratio is majorly impacted by the weather, the link interference plays an important role, and especially the co-channel interference, due to multiple access from satellites to HAPS, or HAPS to RIS-UAV. The interference management directly impacts the buffer sizing at the HAPS that stores data for scheduling transmissions to mitigate interference. Practical dimensioning of the buffer at HAPS is open, yet critical challenge due to the sizing constraints of the SAGIN system components. Motivated by this challenge, we analyze the throughput and buffer size in SAGIN systems, towards a practical solution, based on the varying weather conditions, types of transmission links (FSO/RF) used, and the characteristics of nulling interference from RIS-UAV on the HAPS buffer. We first use a Markov chain model to analyze the throughput, and based on the model provide practical analyses of the resulting HAPS buffer size. Numerical results show the usefulness of the analytical model to deriving practical throughput and buffer sizes for a range of weather conditions and link configurations.