SkyServer Traffic Report “” The First Five Years

The SkyServer is an Internet portal to the Sloan Digital Sky Survey Catalog Archive Server. From 2001 to 2006, there were a million visitors in 3 million sessions generating 170 million Web hits, 16 million ad-hoc SQL queries, and 62 million page views. The site currently averages 35 thousand visitors and 400 thousand sessions per month. The Web and SQL logs are public. We analyzed traffic and sessions by duration, usage pattern, data product, and client type (mortal or bot) over time. The analysis shows (1) the site’s popularity, (2) the educational website that delivered nearly fifty thousand hours of interactive instruction, (3) the relative use of interactive, programmatic, and batch-local access, (4) the success of offering ad-hoc SQL, personal database, and batch job access to scientists as part of the data publication, (5) the continuing interest in “old” datasets, (6) the usage of SQL constructs, and (7) a novel approach of using the corpus of correct SQL queries to suggest similar but correct statements when a user presents an incorrect SQL statement.