Hello Sudhir,
1) there are no fixed figures an old rule of thumb was 50 routers per OSPF area but this is not true anymore.
Scalability is achieved by reducing routing information to a minimum:
OSPF calls for the use of stubby areas or NSSA (not so stubby areas). Using multiple areas allows for route summarization and filtering (of internal routes only )
For OSPF the biggest limit is the number of OSPF speakers on the same IP subnet this causes load on OSPF DR and OSPF BDR
A good rule is to manipulate OSPF priorities so tha different devices act as OSPF DR in different IP subnets to distribute computational load to different devices.
Having 200 routers in the same IP subnet is not a good idea. Having 200 routers in the same area is allowed (in multiple links)
see OSPF design guide
http://www.cisco.com/en/US/tech/tk365/technologies_white_paper09186a0080094e9e.shtml
similarly for EIGRP the use of per interface route summarization (core to edge), of EIGRP stub router feauture can provide the desired scalability
http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080093f07.shtml
2)
OSPF use a cumulative metric adding cost of each link to the destination. cost by default is calculated using the interface bandwidth.
Cost to reach a device on a lan is counted using the BW value on local node to the LAN segment.
There is a reference bandwidth and its value can be tuned to support very high speed links like GE or 10GE
cost of one interface is a 16 bit integer.
inter-area cost is 24 bit integer.
cost = REF BW / interface_BW [ kbps]
the command ip ospf cost in interface mode can be used to assign directly a cost
so if going via 3 links of costs 10, 1, 45 total cost is = 10+1+45 = 56
3)
DDNS = dynamic DNS should mean that DNS entries can be updated by external servers like for example a DHCP server to allow to update automatically a DNS record
Hope to help
Giuseppe
