State of the Art of DBaaS

The folks of Tesora have published an infographic about the market of DBaaS (Database as a Service) one of the fastest growing niches in the Software Industry. What do you think about it? Are you using DBaaS? Should we offer ToroDB on the Cloud as a DBaaS? Which is your favourite Provider?

Please, don’t hesitate to share all your thoughts and ideas with us, using the blog Comments or our Social Networks.


SELECT what, why FROM “8Kdata”;

what why

If anything, we are a group of passionate developers and database enthusiasts.

8Kdata is a database R&D company, aka we-love-databases-and-we-come-up-with-geeky-projects-about-them. Cool thing is that some of those geeky projects end up being serious products that will hopefully revolutionize our industry in one way or another.

We are (and started as) a PostgreSQL company. We love Postgres since we first knew about it, circa year 2,000. Postgres is the most advanced open source database –that’s the motto. And it deserves a companion of high quality Professional Services. That’s what we strive to offer, be it consulting services, support or training.

Back to the products. After researching whether it was possible to create a billion tables in a database, we started ToroDB. ToroDB is the first database that is both NoSQL and SQL at the same time. It speaks the MongoDB wire protocol, so you can use your MongoDB drivers, tools and programs; but data will be transformed into relations and stored in a PostgreSQL database. Which you could query with either the MongoDB API or pure, native SQL.

Then there’s also BigToro (coming soon!), a ToroDB’s offspring. BigToro wants to really enable Data Analytics for NoSQL. The initial advantages of unstructured data turn to really poor results on aggregated queries, the ones used in analytics. By an automatic process, BigToro replicates from a MongoDB cluster, structures data in relational tables, and stores it in PostgreSQL or shards it into analytics/warehousing databases (like Greenplum or CitusDb) to speedup aggregated queries, performed in SQL.

As we completely believe in open source, both ToroDB and BigToro are open source. Fully. We don’t do “open core” and then “enterprise” proprietary software. We’re fully committed to open source and all what we do is open source. Forever.

Since the introduction of the NoSQL databases, users need to choose between either a NoSQL database or a relational one. The NoSQL databases are meant to provide agility and scalability, at the cost of trading either durability, consistency and/or good query capabilities. The relational databases specialize in durability and consistency, but are harder to scale and shard.

Wouldn’t it be possible to have both the benefits of the NoSQL and the relational databases in the same place? This is one of the questions that we repeatedly asked to ourselves.

It’s surprising the number of NoSQL databases that have appeared in the last years. What’s even more surprising, at least for us, is that they both started from scratch and have their own API/protocol. Isn’t that a lot of re-inventing the wheel? In our opinion, it is.

So we asked ourselves a second question: is it possible to create a NoSQL database, without throwing away the experience and know-how acquired by the relational databases during the last 3-4 decades? Is it possible to construct such a NoSQL database based on a relational database, that would also be compatible with existing APIs/languages?

Our answer is, of course, ToroDB.

We believe there is room to create better databases. That despite the myths of relational databases being slow, RDBMSs have proven to be really fast and are software that you can really trust. We believe that in order to have NoSQL you don’t have to sacrifice transactions, strong isolation and consistency semantics and, of course, you don’t have to sacrifice SQL, the lingua franca of data querying.

We also strongly believe that databases have to be fully open source. After all the NSA and spying scandals, no organization (big or small, but specially Governments) should store their data on some software which may be sending data or opening backdoors to spying agencies or other third parties. The only viable alternative is open source.

(1 row)

Back to top