You are aware that 99.999% uptime means something like 5 minutes of actual downtime per year, right? Salesforce is nowhere near that, especially if you consider the many partial outages and disruptions, slowdowns, etc. Heck if you look at trust.salesforce.com right now, you'll see there's only been something like 11 days in the last 30 where there were no reportable incidents across all nodes.
You're going to be hard-pressed to find 99.999% in a SaaS SLA. Even when it is (for instance several online providers "guarantee" 99.95%), you'll find many SaaS providers routinely violate that SLA but your contractual recourse is generally not much better than some partial credit that is based on the amount of downtime.
So, first thing to validate is do you really need 24/7 99.999% uptime? If you are sure that you do, and partial credit for a provider going below that is not acceptable recourse (it wouldn't be if you really need 99.999%), your best option will be to host a data copy yourself, in a physical secure data center that you have total control over. Or in multiple controlled data centers. You could also host in multiple data centers with a co-location provider that has a good track record of security and uptime.
There are quite a few types of workable approaches to shadowing your SFDC data to other data centers. So you could use SFDC as the data front-end for management of data, but for serving it to whatever service you have that has the 99.999% requirement, you go straight from your hosted "data warehouse" that has 99.999%.
Even still, you'll find that managing under 5 minutes of downtime per year is harder than it sounds, even with multiple redundancy. It takes a lot of money and effort to actually achieve.