There is no evidence of Arabian males from the medieval era fathering millions of Somalis. It is so unscientific and ridiculous I won't even bother providing more arguments.
The Arab factor might be far fetched. However, the preliminary TMRCA of HG T Northerners (Isaaq, Samaroon, Ciise) based on STR's does not look that old; it is looking very likely that the founder effect they descend from definitely occurred within the last 1000-1500 years. The Isaaq and Samaroon 111 STR's are almost a perfect match. Even my own 111 STR's which are more differentiated than the 111 STR's of the Isaaq and Samaroon shares an ancestor with them within the past 1300-1700 years. The Y16897 lineage among Somalis does not look as diverse or as old as the carious E-V32 lineages found among Somalis. Nonetheless, I acknowledge that STR's are unreliable when it comes to looking at lineages that are older than 500 years due to mutations and back mutations etc. For instance, from a 111 STR perspective, I was more closely related to Ashkenazi Jews and British Y16897 individuals than I am to many of the Arab samples on FTDNA when I made comparisons using McGee's calculator. The TMRCA's were roughly 2000 and 2400 years respectively.
Once the Big-Y results come in, we should nail down when we separated from our closest Y16897 Middle Eastern/Ashkenazi Jews/Western European relatives. However, two Somali Y16897 individuals are required to obtain an estimate of what our TMRCA is.