Epilogue - Data Storage for Social Networks: A Socially Aware Approach - page 41

such as geographic locations of the servers and how these servers are interconnected
in the server center (e.g., using a DHT). It is thus an interesting problem to
achieve the optimization objectives reckoning with the heterogeneity of cross-server
communication costs.
Social locality is not the only property that should be taken into account in
data storage systems. Locality of other types such as content-based locality and
geographic locality are also important. For example, the importance of geographic
locality is evident in the success of Akamai to improve data access by deploying
content servers closer to the users. One could argue against preservation of social
locality because it may weaken other locality properties such as geographic locality.
However, a nice property of preservation of social locality is that doing so will, to
a large extent, also result in preservation of geographic locality. Indeed, it has been
empirically shown that in an online social network geographic proximity increases
the probability of friendship. Specifically, the number of neighbors of a given
user decreases quickly with geographic distance and, consequently, most neighbors
should stay in the local region of this user. For example, in LiveJournal.com, letting
P.ı/ denote the proportion of links with geographic distance ı, it has been observed
in [ 24 ] that the relationship between friendship and geographic proximity can be
modeled as P.ı/ D 1=ı 1:2 C 5 10 6 . Also, it is reported in [ 12 ] that 58% of
the links in FourSquare.com, 36% in BrightKite.com, and 32% in LiveJournal.com,
three popular OSNs, are less than 100 km; these OSNs have average link distances
of 1,296 km, 2,041 km, and 2,727 km, respectively.
To conclude, social locality should be considered one of the most important
design factors in distributed social data storage and more research in the socially
aware approach is definitely worth pursuing.