This page is from our blog archive
You may still find this page interesting, but there are more relevant and up-to-date blog posts available. Use search or filter options to find the latest content.
Last week, the KeyLines team flew into San Jose, California, to attend and present at NoSQL Now! 2014 – the largest NoSQL forum in the world. It’s an exciting time to be a part of this movement, as the concepts and ideas involved mature from academic exercises to real deployed systems.
There were dozens of speakers, covering 19 separate tracks. Of course, we didn’t get to see them all, but four key messages jumped out at us during the two days.
Use the comments section at the end of this post to share your own experiences.
Our presentation was one of the only ones discussing visualization, and the only one that talked about how to present large volumes of data to business users of applications.
Of the other presentations that mentioned visualization, it was often used as a means of visualizing the database structure or model itself – something that is certainly useful to the engineers and architects designing the application but of limited value to the end-users.
Obviously, we’re biased, but this makes no sense to us. Why dedicate so much time collecting and storing the data, but put so little effort into presenting that data to your users? Visualization should be a core element of your database project, not something tagged on afterward.
In previous years, the NoSQL movement has been about how to get away from the relational database model at all costs. NoSQL was assumed to mean “No SQL”.
This year, though, there were a number of presentations that acknowledged that there is still a lot of value embedded in data stored in relational databases, and depending on the structure of that data, a tabular view of that data may still be the most efficient storage and query mechanism.
Unsurprisingly perhaps, the Oracle team has a presentation on how to work with RDBMS and Hadoop data in the same project.
Over the past couple of years, even the term NoSQL has been adapted to mean “Not Only SQL”. We can certainly back this up – we have implemented KeyLines on top of standard relational databases, and although it requires a bit of data modeling, it can sometimes be the best choice for complex data sets.
There was no shortage of graph database vendors at this conference – evidenced by the panel discussion on selecting the appropriate graph database for your project…
Although both property graph databases and RDF databases are still a relatively small part of the NoSQL world, they are coming into their own. While in previous years, there was an assumption that graph databases were only useful for analyzing social connections on networks like Facebook or LinkedIn, there is growing recognition that graphs are more extensive than that, and that many business problems are best solved by modeling the data as a graph.
We’re not exaggerating. This was the first year that we heard performance metrics that astounded us.
Bryan Thompson from Systap presented a solution that, by using parallel GPUs, could traverse billions of nodes in a millisecond response time.
That level of in-memory graph storage and analytics increases the importance of visualization like KeyLines. After all, your credibility hangs on being able to show your users all this amazing analytics you’ve done server-side.
We had a great time meeting the NoSQL community and vendors. Even though the focus was heavily on server-side storage and analytics, our client-side visualization seemed to strike a chord too.
We’d like to thank the Dataversity / NoSQL Now! team for putting on another successful event. See you in 2015!