". . . if a situation happens like DC isp1 connection down and branch isp2 connection were down."
Ah, understood.
Well, again, with some "work" you can still use your OSPF path, or, what we did in an international company using two MPLS providers, to handle the situation you describe (without usually having a "backdoor" path), we designated certain sites (like your DC and DR) as transit sites. In this case, the branch would use its remaining ISP2 link to get to DR, which would send it to DC across ISP1. (Incidentally, in our situations, every site had two MPLS providers, but not always the same two. So failure of one of the ISP's site connections often led to traffic bouncing off another "transit" site.)
I recall (?) our EMEA region did have two major sites, relatively close together, with their own site-to-site link. I think (?) they just set it up as another BGP transit.
When I mentioned "work", as an example in our case, one of our MPLS international vendors has one (private) AS for their world-wide BGP cloud, the other had regional (private) ASs. (BTW, the latter was better, as in a failure situation, you don't want to use a "transit" site, to jump between providers, that's, literally, on the other side of the world.) As we wanted to use both providers, equally, we had to prepend AS hops on the former to make AS hop count equal to the latter, between sites.
PS:
BTW, again, as (mentioned) our sites didn't have all the same MPLS providers, what you describe as a double failure, breaking the path between Branch and DC, in our case might be caused by a single link failure, on one side or the other. So, bouncing through a "transit" site was critical to maintain connectivity between sites that no longer shared the same MPLS cloud.
