There is good justification for having two IP addresses on a PLC.
Yes, PLC should not be on the corporate network. We have a business (corporate) network, and a production network - completely isolated.
We like to have the PLC, HMIs, & Drives on the production network AND THEN all the remote I/O, intelligent OLs and such on the local 192.168.1.X so we can use the thumbwheel to set the address on the IO. This easy allows replacement for IO components WITHOUT the use of Rockwells wonderful BOOT-P at 2am.
This allows us to login into the PLC & Drive for programming & troubleshooting,
It also gives the PLC better access time to the IO since the multiple drives and HMI are not hogging the bandwidth. Even if you use an EXPENSIVE Level 3 switch all traffic is still going through on cable into one port. But I do like the idea of being able to ping the components on the local subnet like you can do with a Level 3 switch like "The PLC Kid" does.