Comment by didgetmaster | Hacker News Reader

Comment by didgetmaster | Hacker News Reader

The test results are impressive (assuming that they are not just cherry-picked) against the competition; but I seriously wonder if anyone really cares. People will tell you all day that performance is super important, but in my experience they rarely 'put their money where their mouth is'.

For example, I created a new kind of relational database engine that also has some very favorable performance characteristics when compared with popular engines like SQLite, MySQL, PostgreSQL, and SQL Sever. I have posted some videos (https://www.youtube.com/watch?v=Va5ZqfwQXWI and https://www.youtube.com/watch?v=OVICKCkWMZE) and done a number of live demonstrations of my system showing how it is many times faster with both transactional and analytic operations across a wide variety of queries.

I certainly didn't expect mass adoption where everyone decided to dump their old system and move to mine (it is still missing some important features). But in a world where even a 50% increase in speed should be seen as significant; the reaction has instead been crickets! You would think that showing a 10x improvement at anything would result in some serious inquiry by at least a few hundred people. Instead, I seriously wonder if anyone cares about speed anymore.

Do you have an open source repo? Commercial licensing? Is there a business / pricing model? Do you have a roadmap for feature parity with other solutions? Do you have a public list of the missing features? Is there written documentation? Can the db server be interacted with aside from the Browser app?

I don't mean to downplay the potential performance benefits you've shown, but you're up against other databases that have been around for 22-33 years each. You have a bus factor of seemingly 1. You have a specific way of communicating that you've decided on that is unlike any other popular solution out there. I wouldn't assume speed is the primary factor.

You are right about this 100% percent, performance is not the only factor, especially in an established DB space such as a relational DB world. There is a lot of things to consider before moving/or picking the database.

I totally agree. Just like if a new car came on the market which claimed to have an engine that gets 100 mpg and wasn't the size of a golf cart; there are many other factors that go into choosing a car (comfort, quietness, safety, price, etc.).

But you can bet if such a car was announced, a whole bunch of people would be lined up to take it for a test drive to see if it really got the claimed mileage. If it met those claims, you can also bet that investors would line up to help iron out any kinks in the system that might keep customers from buying it.

Why is this not true for data engines as well?

Didgets is currently in beta with a free download available from the website https://www.Didgets.com so anyone can try it out with their own data set. I haven't decided yet what parts I am going to open source. The engine has an API so many other programs besides the browser app could interact with it.

It is a little startup with a tiny 'bus factor' so like I said, I don't expect everyone to just jump on board; but I expected to get a little more attention from the 'curiosity crowd' who just might wonder why it is so fast. It needs more resources to support other platforms (the code is cross-platform, but I have only tested on Windows and Linux so far) and languages.

Again, the browser is a proof of concept app that demonstrates that the underlying engine is not just blowing smoke. If it can't attract some interested customers and/or investors, then it will be a dead-end project just like every other promising technology that failed to get a hold in the market.

So these results were not cherry-picked since the same queries were run previously in our CI infrastructure. Yep, as you have said, performance is just one of the things that matter. A lot of things matter when picking a vendor, some are mentioned in the comments.

What is specific in graph space is that things are still quite early days compared to relational database space. This means performance differences are big, and playing a more important role.

Performance in graph databases is critical, and would need improvements of several orders of magnitude, rather than tens of percent points. Many algorithms are impractical in real world graphs because they are often O(n^2) or more, and in practice we have to resort to approximated algorithms instead (like pagerank and all the likes).

Reply on news.ycombinator.com