Currently, when an unhealthy VDC is detected while geo-pinning, the client changes the operand in the mod operation to reflect the number of healthy or good VDC. This new healthy VDC count is used to compute the geo-pinned VDC for every request versus just handling the bad VDC-pinned objects. Computing a new VDC just for the bad VDC requests would guarantee less data placed in the "wrong" VDC.