From 2012, I was following the thumb rule of maximum 50 routers in an area for My OSPF network ( which consists of more than 500 routers and 15 areas). At that time, the major number of routers were 1800, 2800, 3800 series. But now network is having 1900, 2900 routers and again we are replacing with new ASR and ISR routers. Considering the increased computer power of new devices, is't still valid to follow the same practice which created 7 - 8 years before? Or is't changed ?? I am unable to find it now, so please help..
on a side note, they size of an OSPF area used to depend on limited CPU and memory resoiurces, which nowadays is hardly a factor anymore. There are a lot of other factors to consider. Have a look at the article below, I think it sums it up well:
AS Georg's reference notes, this is very much a "it depends" kind of answer.
There are important considerations that are not mentioned in Georg's reference, such as a particular vendor's OSPF implementation. For example, Cisco has done much to insure OSPF stability while still meeting OSPF standards. I've seen brand "X" routers go into OSPF meltdown in situations that didn't bother Cisco devices.
Although memory is much, much more on devices, CPU can be limited by how busy the router is. On small software based only routers, as the CPU's load often goes hand-in-hand with transit traffic volume, a very "busy" router won't have excess CPU cycles to spare, which are more likely to be needed with "too many" routers within an OSPF area.
"Powerful" L3 switches, somewhat surprisingly, often don't have much general CPU "horsepower" since the bulk of their transit traffic is processed by dedicated hardware. So they too can be found to be crushed by a too large OSPF area.
When an OSPF area is stable, it often can be very large, but just one flapping link can cause havoc. So, another consideration is how else you configure you devices, for example using dampening might allow you to "safely" run with a larger OSPF area.
Basically, OSPF area size designs, and the number of routers within them, have much to do with how "disaster proof" (with its cost) it is.
there are other design considerations that you need to take in account:
All OSPF route manipulation can be performed only for internal routes and only at area border by ABR nodes.
So if you want to build a scalable network with the capability to control route propagation you need to use a multi area OSPF design with area 0 on the center.
So avoid use of easy shortcuts like redistribute connected, because they create unnecessary external routes that you cannot have fine control over them (OSPF allows to use stub areas or NSSA areas but you can only choice between allowing all external routes to enter a standard area from backbone area or to stop all of them if using a stub or NSSA area type).
So actually how many routers you can put in a single area is a good question for specific scenarios like MPLS traffic engineering, but generally speaking using multiple OSPF areas is in most cases a wise choice for the added benefit of route control (you can filter or summarize internal routes at area borders with fine control using prefix-lists for filtering) in addition to separation of failure domains.
OSPF is link state in nature and distribute-list only can avoid to install a route in local IP routing table, but they do not stop LSA flooding within a single area.
For this reason I recommend a multi OSPF area design from beginning even in a "small" network.
Hope to help
Its all depends on how big your OSPF area, suggestion is make it less as possible you can and make different area, so it is easy for diagnosis the issue, when the incidents happens.