Now Cisco has produced this magnificent compute engine, coupled with this equally excellent Data Centre grade switching family both of which seem to be maturing, I am struggling - repeatedly - to convince myself that not only is near complete virtualisation is desirable it is in fact inevitable and I should be making steps towards planning this today.
Can anyone share their experiences with the CSR1000v, the success - failures and recommendation?
If from a cloud providers point of view, they can provide multi-tenanted environments - then from an enterprise point of view why can I not simply leverage this same approach to logical division for my own trusted/semi-trusted/untrusted network segments - why should I not basically put as many services as I possibly can into a virtual environment and provide the bare basics for my own store/transport/compute requirements and present an ethernet hand off to everything else? Let my carrier deal with the variability in the circuit type, and connect to him with ethernet leaving me with a near completely uniform architecture that simply scales up as I need it? More horsepower? New blade, new chassis, new filer, new core? Bolt on, snap off. etc. In the SDN environment is there really a tool that is 'overkill'? Can it just not be scaled back or up depending on my needs, which are likely to change dependant on my companies business cycles?
Is there any strong arguments against this kind of approach now?
