Global architecture advice

[Don Benkendorf]

Hello,
I have a very large (15000+ managed nodes) global environment I'd like to manage with HP SIM. NOTE: We're also being demoed with HP ICE later this month, by an HP rep, so that may come into play, later. I've read the PDF file "Configuring HP SIM for optimal performance in a large environment", looked at the HP Sizer tool, and many of the other available docs. It seems like my environment is larger than what's dealt with, commonly. Also, my environment is quite heterogeneous, although almost completely HP hardware (mostly ProLiant, with some Integrity) : HP-UX 11i v1-3, with a few 11.0 and 10.2 boxes, Windows (most server flavors), Linux (mostly SUSE 9 and 10) and the odd AIX and Solaris boxes. We also have VMWare hosting of Windows on Linux (typically) and ServiceGuard installations. We're an Oracle database shop, with some MS SQL Server 2005, only where required. Question is, how to best architect this from the top down ? How many CMS's and Oracle repositories, located where, on what platforms, managing what mix of management domain nodes ? Would obviously like to have as central a point of control/access, as possible, to the whole works. We have 3 data centers, 2 in the U.S. and 1 in Asia Pac, which would most likely be where the CMS's and DB's would be. Best to have Windows-based CMS managing Windows nodes ? Best having Linux-based CMS for managing Linux, HP-UX and "other Unix" nodes ? Hierarchical CMS architecture with one "main" at top doing polling of ALL nodes (can any h/w, CMS combo handle 15,000+ nodes ?), with regional CMS's doing event management ? Please, can anyone point me in the right direction, for this - docs, etc. -or- provide some advice. Many thanks, in advance. -- Don Benkendorf