domain driven design - Normalized or Denormalized Data in Microservices and Service Composition -
so our development team has been working towards microservices past 6-8 months , have picked lot of steam.
we have experienced several gotcha
moments in time, , humble enough know in many more move closer moving our platform production.
one area can't quite put finger on how treat our data between our service boundaries. hear lot of statements form large companies have implemented microservices, can never seem straight advice , reasoning.
specifically, given 2 service domains user
, contacts
,and assuming user
has contact
object associated it, options each of these 2 service domains in regards managing own data?
should user
have contactid
stored it, or should store entire contact
object?
i have seen many reliable service oriented development teams (netflix,amazon,nike,etc) make statements such following:
"normalization root of evil..."
"break shared..."
"share nothing..."
well, statements wrong. modularisation finding right combination of low coupling , high cohesion context. user can store copy of contact object long aware snapshot of @ moment in time. many use cases fine, , need ensure work real contact known responsible domain.
Comments
Post a Comment