ravendb shardstrategy & shardeddocumentstore -

-

can choose have few objects sharded, typically have potential have large amount of data , leave others stored across instances?

for example when have banking transaction, might want store customers , transactions in different shards , might want store list of bank branches in databases. possible?

you can that, not in single database. way works, have 2 databases on each server (and 2 document stores) 1) sharded, transactions & customers 2) standard, banks.

the banks database replicated nodes.


Comments

Post a Comment

Popular posts from this blog

php - Global connection variable or new one every time -

-
i'm thinking of creating global variable $pdo use along script inside classes , functions , code . there security breach can occur upon ? i'm using seperate connection every time . shall stick or use global pdo object ? this depends on application, there cases want manage multiple connections. however, in many cases, makes sense open single connection use during course of processing single request. helps eliminate overhead of opening/closing connections throughout code. guess security not of concern in cases unless planning on handing off db connection set of code should not have db permissions associated specific connection (maybe library or service not control). i highly suggest concept of dependency injection, in cases preferable pass around instantiated db object pieces of code need rather treat global , use global keyword access it. helps avoid needing replicate code in every class/function check db has connection , such. if have single class , can ...
Read more

c++ - protocol buffers - generate NON-inline accessors -

-
we're using protocol buffers (2.4.1) in medium size embedded system w/ c# c++ code. use protobufs isolate our managed , native layers w/ easy maintain serialization layer (for curious, have used pinvoke, have run native code in separate process on test/simulators). our system has number of dll's, , have generated native protobuf code in it's own dll other parts of system can don't have link in generated code directly. the problem i'm having generated accessors inline , eg: inline const ::myprotoclassname::myfield& myprotoclassname::myfield() const { return myfield_ != null ? *myfield_ : *default_instance_->myfield_; } use generated api in size (the ' default_instance_ ' dereferenced , accessed if particular field not set). means can't link (lnk2001) clients using accessors because there not symbol default_instance_ i think typical use case protobufs have each component link in generated protobuf code (after all, serialization...
Read more