Open cbenoist opened 12 years ago
I know we are looking at this to host the solution while working, but I'm curious what the plans might look like moving forward if we decide to blow this out beyond a few users. Would we still use Azure or perhaps go for some free storage location to run it from if this project become more than a test/research effort? Trying to spark some conversation about options.
Good point. Initially my idea was to utilize our free Azure SQL storage and bandwidth, but that could get quickly out of hand. Given the cost considerations, it may not be our solution when we expand the rollout. I am open to ideas of about using alternate storage formats. doyou have anything in mind?
I wasn't thinking of anything in particular, just that we need to be open and flexible to support various data storage mechanisms. I would be interesting to make the design of the data hooks generic so we can plug, DI, other providers in place easily.
Good point. The model and the data interfaces we design will not be implementation specific. Then our implementation will abstract all of the specifics, so we can change the data store as needed. One thing to keep in mind is, I have no experience with NOSQL or Document DBs My model design thinking is based on a Relational model because that is what I am used to. Would be interesting when we look at Mongo or Raven DBs.
-----Original Message----- From: Cliff Benoist [mailto:reply@reply.github.com] Sent: Wednesday, February 29, 2012 2:30 PM To: Sivakumar,Raj Subject: Re: [Slech] SQL Azure backend (#3)
I wasn't thinking of anything in particular, just that we need to be open and flexible to support various data storage mechanisms. I would be interesting to make the design of the data hooks generic so we can plug, DI, other providers in place easily.
Reply to this email directly or view it on GitHub: https://github.com/cbenoist/Slech/issues/3#issuecomment-4244887
I don't think that we should think about models in terms of what the implementation looks like in a storage target. I think that if we focus on the relationship between the model objects then we have it made. We can talk about the modeling project I added to source code and really try and get at what the relationship should look like between our objects. Azure as a testing data store is a good idea, as long as our implementation doesn't get chatty.
Shelving this for now and focusing on no SQL databases for learning and implementation.
Look at SQL Azure as a possible backend for the system. Report back findings.