In general, if you've experiencing "latency", flow control shouldn't help unless the "latency" is due receiver packet drops and even then, the transmitting device may drop the packets instead, possible even more so. I.e., flow control is going to force more packets into the transmitting device's port queue. Basically, flow control should keep the Dell SAN device from dropping packets it receives, which it cannot keep up with, but in turn, those packets are now queued on the 9K and can drop there. It's possible the 9K might, without tuning, drop more packets than the Dell SAN device would with FC disabled.
If packets are being queued and/or dropped, you want to determine where's that's happening and if that might be mitigated. For example, perhaps the problem is the SAN sending to the client, which I assume (the client) does not have a 10g connection.
In answer to your question, no, you wouldn't want to generally enable Ethernet flow control end-to-end (or generally use it), especially the "ordinary" Ethernet variant which completely blocks the port. Again, it's generally better if you try to eliminate any packet drops and/or better manage port congestion using QoS techniques.
Consider if there's congestion on the edge<>core link and you have a mix of traffic. E.g. traffic to/from the SAN and perhaps VoIP traffic. If the SAN traffic is causing congestion, you don't want to block the VoIP traffic.