You aren't allowed to install anything (whether Oracle or otherwise) on the storage cells. Hence why using the EMCC plugin is the preferred monitoring method (it doesn't install anything on the cells either). Without using the plugin, you can gather oswatcher archives if you wish. For reactive problems, you should configure alerting from the storage cell (via email or SNMP). I'd highly suggest using the EM CC plugin as that is where Oracle is spending its efforts to make storage cell monitoring easy.
the supported and recommended method is enterprise manager. Tehre is a special standalone version you can install for Exadata if you do not want to have a full fledged setup.
Apart from this some people use collectl but it is not endorsed by oracle.
okay thanks for the answers. :-)
Hopefully we will fix the problem with the em cloud control agent so that we can use it in a few month.
In the meantime we use for the cell server monitoring some command line statements inside cellcli like
cellcli -e list alerthistory detail
or something like this especially for disk predictive failures.
CellCLI> LIST PHYSICALDISK WHERE diskType=HardDisk AND status="predictive failure" DETAIL
CellCLI> LIST PHYSICALDISK WHERE diskType=HardDisk DETAIL;
May be we are also in considerations for using the "sundiag.sh" script.
But here I am not sure if it makes sense.