NoSQL and CQRS

[linus.dm]

It seems that CQRS is a perfect fit for some of the new NoSQL
solutions that are gaining interest these days. I'm not sure though
where to use these. I can imagine using a Key-Value solution to store
all events in an event sourcing system. But would it equally well fit
the needs of the read-only components in a CQRS systems?

Has anyone played around with this combination of NoSQL and CQRS? What
were your conclusions? What are your overall thoughts on this?

Thanks in advance!

[Rinat Abdullin]

Hi,

My favorite one at the moment is about using SQLite database stored in Azure BLOB, that stores indexes, and some schema-less values serialized by ProtoBuf.

This worked around slow SQL Azure and Table storage (lag problems), allowing to sync in multiple millions of products with arbitrary metadata per customer in matter of seconds. Random reads are really fast (esp. when SQLite file is cached locally).

Slightly more details: http://stackoverflow.com/questions/3231809/3234050#3234050

We are also researching into content-based storage (hashed, like git) and actively using append-only event streams (http://abdullin.com/journal/2010/6/3/time-machines-should-support-linq.html)

It seems to be fun all the time))

Best regards,
Rinat Abdullin

Technology - Lokad.com | Blog - Abdullin.com | +7 917 4613826

[Adam Dymitruk]

This is what I'm doing for event streams, git-like content addressable add only system.

Adam