disable data synk option in record type

The impact of it is huge. I won't advise doing that unless that is a hard requirement and you don't plan to go back. As David mentioned, you will need to update all the record references at every single place where it is being used. 

suppose there is  project and in the record type we can view 4 million data and also what you want to do with those records, you have to do only with those limited 4 million records ,but if  I  want to perform some action beyond 4million record ,then how you will configure that ?

I think your answer will be source filters but if there is an requirement that we can not apply any source filter then how you will configure this ?

And also is it recommended to use CDT instead of record type ? If Yes , then how you will build the relationship between two CDT and access the second CDT data using first CDT .

suppose there is  project and in the record type we can view 4 million data and also what you want to do with those records, you have to do only with those limited 4 million records ,but if  I  want to perform some action beyond 4million record ,then how you will configure that ?

I think your answer will be source filters but if there is an requirement that we can not apply any source filter then how you will configure this ?

And also is it recommended to use CDT instead of record type ? If Yes , then how you will build the relationship between two CDT and access the second CDT data using first CDT .

A non-synced record can easily work with more than 4 million rows. But that means to sacrifice all the features sync brings along.

It is recommended to use Synced Record Types whenever you can. The main drawback of CDTs is that there is no versioning and no security that is applied to them.

You can only use relationships on synced data Records.

For CDTs, you can achieve that via nesting but it is not recommended (for performance reasons). If you are using flat CDTs, your only option is to use multiple queries to build your full object.

but I need to implement the relationship and use that relationship via CDT which is my requirement ,ignoring the performance reasons.

based on my use case I need to work on more than 4 million records so I cant use record type data sync approach !

FYI 

https://docs.appian.com/suite/help/23.4/cdt_design_guidance.html#nested-cdts

try with nested cdt's

Then you can simply use CDTs like you said. Are you facing any issues while doing it?

For me it's not clear what your actual requirement is.

Are you aware that in 24.1 you are able to sync the latest 4 million rows in the database?

https://docs.appian.com/suite/help/24.1/Appian_Release_Notes.html#keep-record-data-available-at-high-volumes