I'll offer some resources:
- A session from last Google NEXT, where Yahoo, NY Times, and Blue Apron detail their experience with migrating to BigQuery [0]
- A session from previous Google NEXT, where Kabam Games detailed their experience with migrating to BigQuery (and running the rest of their workload on AWS) [1]
- A blog "15 things you probably didn't know about BigQuery" [2]
- A blog detailing what BigQuery means by "fully managed" [3]
- A blog offering some details on most recent iteration of Dremel "In-memory query execution in Google BigQuery" [4]
- A blog describing Capacitor, the opinionated storage format that replaced ColumnIO (on which Parquet is based) [5]
- "BigQuery Under The Hood" [6]
- Since you specifically asked about Load Performance, [9]
( A wholly redundant disclosure that I work at Google Cloud)
[0] https://www.youtube.com/watch?v=TLpfGaYWshw
[1] https://www.youtube.com/watch?v=6Nv18xmJirs
[2] https://medium.com/google-cloud/15-awesome-things-you-probab...
[3] https://cloud.google.com/blog/big-data/2016/08/google-bigque...
[4] https://cloud.google.com/blog/big-data/2016/08/in-memory-que...
[5] https://cloud.google.com/blog/big-data/2016/04/inside-capaci...
[6] https://cloud.google.com/blog/big-data/2016/01/bigquery-unde...
[7] https://cloud.google.com/customers/sharethis/
[8] https://storify.com/felipehoffa/dow-jones-on-bigquery
[9] https://medium.com/google-cloud/paying-it-forward-how-bigque...