NoSQL databases - good candidates for log processing/aggregation and rollup? [closed]

As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance. 开发者_开发问答 Closed 10 years ago.

I have a MS SQL database that's used to capture bandwidth stats. We have a raw data table and to improve reporting speed at different drill-down levels we aggregate and rollup data on an hourly, daily and weekly basis to separate tables.

Would a NoSQL database such as Mongo or Raven be a good candidate for this type of application?

---

Different NoSQL solutions solve different problems for different uses - so first off the best thing to do is look at your problem and break it down

• You are writing heavily to storage, therefore write speed is important to you
• You want to perform aggregation operations on that data and have the results of that easily queryable
• Read speed isn't that important from the sound of things, at least not in an "web application has to be really responsive for millions of people" kind of way
• I don't know if you need dynamic queries or not

Let's look at Couch, Mongo and Raven in a very high level, generalised way

Raven

• Fast writes
• Fast queries (eventually consistent, pre-computed, aggregation via map/reduce)
• Dynamic queries possible, but not really appropriate to your use case, as you're most likely going to be querying by date etc

Mongo

• Blindingly Fast writes (In my opinion dangerously, because power going off means losing data ;-))
• Slow reads (relatively), aggregation via map/reduce, not pre-computed
• Dynamic queries are just what_you_do, but you probably have to define indexes on your columns if you want any sort of performance on this sort of data

Couch

• Fast writes
• Fast-ish reads (Pre-computed, but updated only when you read (IIRC)
• Dynamic queries not possible, all pre-defined via map or map/reduce functions

So, basically - do you need dynamic queries over this sort of data? Is the read speed incredibly important to you? If you need dynamic queries then you'll want Raven or Mongo (For this sort of thing Couch is probably not what you are looking for anyway).

FWIW, Mongo's only use case in my opinion IS for logging, so you might have an anwer there.