I admit I dunno how I got here in the first place … ah, right, yesterday was Paddy’s day and I was sitting at home with a sick child. Now, I tinkered around a bit with a hosted CouchDB solution to store/query JSON output from a side-project of mine.
Then I thought: where are we re hosted NoSQL in general. Seems others had that questions as well. So I sat down and here is a (naturally incomplete) list of so called NoSQL datastores that are available ‘in the cloud’. Most of them with an established freemium model, few of them in (public) beta. In terms of type (K/V, wide-column, doc, graph) we find quite everything, incl. proprietary types – like Amazon and Google have – where it’s sorta hard to tell what kind of beasts they are. Not that it matters, but for completeness
OK, nuff time wasted, here we go:
Amazon’s hosted NoSQL datastores
Both SimpleDB and DynamoDB are sorta key-value stores where the latter seems to be for more serious business (scale out). They explain the difference between SimpleDB and DynamoDB in detail. Pricing is in place, looks sensible. I have not tried any of these yet.
Google’s hosted NoSQL datastores
Tightly integrated with Google App Engine (GAE) comes the datastore with its own query language. If you’re on GAE, this is what you get and what you have to use, anyways. And then, since a bit more than a year there is BigQuery with which I’ve been toying around now for a year or so. Very performant and powerful but not the most obvious and clear pricing strategy.
Joyent offers a so called Riak Smartmachine. I have been toying around with Riak a while ago but haven’t found time to test Joyent’s Riak offering (though I’m pleased with their Node.js offering, hence assuming similar level of service, documentation, etc.).
Cassandra in the cloud
I only found one hosted Cassandra offering. Can that be? Didn’t look closer. Anyone?
So, both cloudno.de and Cloudant offer hosted CouchDB instances (the former also offers Redis). I am currently using the free plan (‘Oxygen’) with Cloudant and find it very straight-forward and easy to use. Prizing looks OK in both cases though I sometimes find it hard to pick the ‘best fit’ for a given workload. Could anyone write an app please that does this for me?
Also for MongoDB I was able to spot two offerings: MongoHQ seems somewhat to be the established player in the field, nice docs and sensible princing. Apparently, Joyent is also offering a MongoDB Smartmachine – anyone tried it?
There are quite some offerings in this area: the general-purpose sort of graph data stores and the RDF-focused ones. In the former category there is Neo4j’s Heroku add-on which I had the pleasure to test drive and found it very useable and useful. And then there is an OrientDB-based offering called Nuvolabase; I have signed up and tried it out some weeks ago and I must say I really like it. Disclaimer: I know the main person behind OrientDB as we’ve done a joint (research) project some years ago.
Last but not least: RDF-focused graph datastores in the cloud. I guess my absolute favourite still is Dydra which I’ve been using manually (SPARQL endpoint, curl, etc.) and in programmatically, inapplications. I think they are still in beta and pricing is not yet announced. And then there is the good old Talis Platform, the established cloud-RDF-store for a couple of years now. Any plans known?
Filed under: Big Data, Cloud Computing, NoSQL