Hacker News Reader

Comment by karterk | Hacker News Reader

It depends on the size of the total addressable market. If the size of your target market is big, even if there are bigger established players, there is always a chance for a small, nimble product to carve out a niche. Established players often head upmarket (especially if they are VC funded). This often gives an opening for a new entrant to capture some of the SMB market share.

An example closer to me: the search engine market is dominated by Elasticsearch, which is open source, extremely popular and is now even a public company. However, I decided to work on a small project called Typesense (https://github.com/typesense/typesense) to try and carve out a niche (and perhaps charge for it later).

I decided to take a dive simply because of the size of the market -- search is required in almost every SaaS application, and there are plenty of downsides to ES that can be addressed with a nimble, alternative search engine. And that's what I am doing :)

View on HN

I'm one of the people who'd definitely use an alternative to ElasticSearch! I hate the fact that elastic search is using Java. It just takes up too much memory.

There are few Rust based projects now for Full Text Search. So you can move away from JVM.

- https://github.com/valeriansaliou/sonic

- https://github.com/toshi-search/Toshi

Great, I would love to hear your feedback on Typesense then. It's not meant to be a full-blown ES replacement. Elasticsearch is a beast but most people don't need that (atleast to begin with).

You could use Lucene directly in a couple of languages:

Python Port: https://lucene.apache.org/pylucene/index.html

.NET Port: https://lucenenet.apache.org/

Please also check Xapian [1].

[1] https://xapian.org/

FTS in PostgreSQL is pretty good, except for missing rank algorithms like tf-idf.

Interesting idea. How's it going so far? Do you have any customers?

Started this as an open source project with the goal of first creating an ecosystem. So far, I have been getting good reviews from people who are using it as a search solution.

Eventually, there will be certain premium features that maybe only 10% of people will really want, and I will charge for that. Playing the long game here.

How do you protect yourself from Amazon offering it on their cloud? Same issue as MongoDB and Elasticsearch.

Good question, and something that's always on the minds of open core companies these days. In my case, I don't mind, and in fact would welcome it! It would solve my no.1 pain point: discoverability.

I'm not planning to build a billion dollar startup out of Typesense, so I don't face the same problems that Mongo and ES do. I also don't plan to offer a hosted version of Typesense. If somebody else fills that void, I will happy to let them do that. For an infrastructure product, there will always be people who want to run it in-house -- there is enough $$$ to be made in that space.

Surprising, AWS is quite open with startups about potential cannibalization if you can form a relationship with the Accelerate program.

Reply on news.ycombinator.com