The data model which we are using right now has one to many relationship between products and sku's that is one product can have many sku's but sku's cannot be common across various products.
Now these products can further be associated with dummy products e.g. suits (Collection of coats,trousers and waist coats). These dummy products and normal products have many to many relationship that is one dummy product can have many products associated with it and also same product can be associated with many Dummy products.
The problem arises when , we have defined the sku id as the record spec, hence when we run the baseline update, the record formed for the suits has all the dummy products having same sku id together which we don't want. I want in case of these dummy products I should get individual records based on dummy product id for dummy product alone.
It sounds like what you will want to do is instead of using the sku_id only as the record spec you should create a new property by concatenating the product_id and the sku_id and use that as the record spec. This should form the true uniqueness of your records.