First steps to using Elasticsearch?

I have a WordPress site that is getting up there in content, and we are looking at related posts systems that don’t seem to work as well. So I am looking at Elasticsearch as a search and related posts replacement. But ES is kinda complicated, so I have some questions.

  • Are three nodes (servers) required? Is there a way to start with one and scale up, or should one start at three?
  • Is it easier to build a cluster, or just use a SaaS? What is the determining factor?
  • How does security work? I keep seeing the commercial security X-pack being mentioned, but is there a non-subscription based way to secure read/write access? Is it IP whitelisting?
  • Is it possible to run a cluster for multiple indices/sites, or does each index create more overhead that make it prohibitive?

If you have any experience with this, please share with me your hard lessons! :smile:

Are three nodes (servers) required? Is there a way to start with one and scale up, or should one start at three?

no, you can have a single-node cluster. and you can scale up. if you’re just testing/developing for now, you can start with one. you can have 2-node cluster for simple failover, c.f. Add Failover | Elasticsearch: The Definitive Guide [2.x] | Elastic

Is it easier to build a cluster, or just use a SaaS? What is the determining factor?

I don’t know, I’m a hobbyist, and I don’t know a lot about your deployment. Are you already running auxilliary systems on your own, e.g. memcache or squid/varnish? It shouldn’t be any harder. I feel like the determining factor between diy and saas is your appetite for control, hassle, and budget…

Is it possible to run a cluster for multiple indices/sites, or does each index create more overhead that make it prohibitive?

Totally guessing based on my very limited elasticsearch experience and some google searches/quick reading about things like ElasticPress, it should be possible to run a cluster for multiple sites. It might be harder to figure out which site is using all the resources, and to have usage-based billing if that matters for you.

1 Like

We created new indices as we needed them. Some held tons of long documents while others had few short docs. As far as I know, we were not concerned with overhead about multiple indices.

1 Like