After 3 drafts of my proposal, due tomorrow on the 6 month marker of my Ph.D study and referred to internally as the Registration milestone I have some minor amendments to make this evening and I am good to go! I will though have to find out who my ancillary supervisors are and meet them and also pass my proposal to another professor, who may or may not end up as an ancillary. So I expect to make a few more changes after based on their feedback but at this point I am relieved as my Director of Studies has donned his Panama hat (Del Monte style) in regards to my proposal and even passed compliment on my progress, a 100 fold! Which either means I am better than better or I was nothing but a wandering fart when I started, I’ll opt for the former.
i) Replace multiple failover points with on-demand instances of systems based on their last known good configuration and ii) gracefully terminate and restart services and systems based on demand and performance to maintain acceptable rates of availability throughout an SoS.
The problems these are countering come from the the nature of a SoS compared to a traditional system. i) In traditional systems, such as the enterprise failure is everywhere and so redundancy is everywhere. When SoSs inherited this virtue of design where there would be one or two failover points in a system, in an SoS there are dozens, and dozens. In my view this has a massive impact on performance and not to mention the climate! ii) In a traditional system the owners of the systems; the IT departments schedule processes overnight in order to maintain reliability. In an SoS there is no overnight, demand is dynamic and so should be the processes that are enacted to ensure a service or system that the SoS relies on runs smoothly.
I suppose I should’ve marked out the problems first then the solutions but hey, it’s too late now as it’s time to go home.