Were you on M0 or M2 shared clusters? I think those were down, while other dedicated instances stayed up.