On Aug 7, 2008, at 11:45 AM, Garrett D'Amore wrote: > All that said, I still believe that existing rstat or SNMP may > provide a better choice to inventing a new API and protocol. But I > suppose that may depend on the stats needed, which is why I asked > for the project team to supply that information. > > -- Garrett
I am unable to supply a detailed list of specific statistics required since my particular work is driven by requests for observability tools from diverse communities ( commercial and HPC to name two ). I have been tasked in the past with providing network, disk, bus, CPU and memory observability tools which were based in whole or in part on data supplied through the kstat interface. I have no problem with teaching rpc.rstat or SNMP new tricks, except that it is difficult for me to anticipate what data may be interesting to consumers. -ejo --- Erik.Oshaughnessy at sun.com, x64070, 512-401-1070, SAE Austin
