1. the CPU/Network/Disk usage you get with "get-hpcmetricvalue" are not realtime, the default interval for each metric is "1min". you can export-hpcmetric to see the interval setting for one metric, change the interval in exported xml file, then change this setting with import-hpcmetric command. however, the minimun interval is 1min, you cannot set this to a value less than 1min.
2. if you are an cluster admin, you can query realtime perfmance counter values with .NET api. pls. refer to
http://msdn.microsoft.com/en-us/library/system.diagnostics.performancecountermanager.performancecountermanager(vs.80).aspx for more information.