I notice that there are apparent problems with the F@H stats server again today. I am seeing a trend here - on most holidays, Stanford has problems with the F@H stats. My experience leads me to the following conjecture:
  • The stats services require a lot of manual intervention to keep running; thus when people are off on holiday, there are interruptions in service
  • The stats services are poorly architected
  • The stats services are poorly designed and/or implemented
  • The stats infrastructure is not adequately robust (lacking capacity, dependability, scale, etc.)
  • Managers of the project view availability of stats services as a low priority
  • Poor management of the stats services contributes to lower confidence in the project as a whole

It's a shame that a project of this size and importance suffers this kind of visible, poor performance. Stats are not beta. They are an important part of the production environment. Stanford needs to get their act together and improve the availablity of the stats services. Perhaps if more top-producing teams complain about it, they would be more inclined to improve the availablity of their reporting.