This was caused by the datacenter switchover on that day (T287539). Apparently the hardware in eqiad is slower.

It has increased identically on desktop, you can find the general data here: https://grafana.wikimedia.org/d/000000085/save-timing?orgId=1&from=now-30d&to=now

If you're concerned about save timings in general, the Performance team has been working on improving it recently (T277788). I don't think there's anything for us to do right now, though.