New cool features in Solr that we can use with Sitecore

New cool features in Solr (6 & 7) that we can use with Sitecore

Solr 7 is finally out! What new features do we have in Solr as part of the last major releases, and what can be benefit from them as part of a Sitecore setup?

That’s what I am going to cover and try out in a number of blog posts. Let’s look at some of the features that looks interesting from the last major releases – Solr 6 and Solr 7.

Replica Types

VERY interesting! And this feature looks awesome. I often get questions around Sitecore and Solr versions – and SolrCould – and this features seems to give us something very useful.

There are a lot of good things about SolrCloud compared to Master/Slave setups. But one should be aware that with SolrCloud indexing happens on all the replicas. This means harder to control performance also on the nodes that handles the front-end/CD request from Sitecore.

On the Master/Slave setup only on the Master handles the indexing. In a Sitecore setup with multiple instances the Master/Slave setup fits well for most situations; Changes happen at the CMS and on Solr Master and get replicated to the Solr Slaves. On the other hand; SolrCloud can scale nicely when indexing goes up (Do I have to say Sitecore Analytics Index?). So, an obvious question would be: Can’t we have both?

The answer seems to be NOW YOU CAN!

A new replica type “PULL” has been added as part of the Solr 7 release. This works pretty much as the Master/Slave replication. A PULL replica only gets the updated with new Index Files, but not the transaction logs. That means that it cannot be elected a leader, and it doesn’t spend time/cpu on indexing. And that’s exactly what we want.

This looks like a good setup for a high performance Sitecore/Solr setup; Having s SolrCloud cluster handling indexing and several PULL replicas for handle request from the from Sitecore front servers.

Read my post on this here

LTR (Solr Learning to Rank)

A feature for extract features inside Solr for use for training in a machine learning model. The model can then be deployed to Solr and used to re-rank your search results.

Solr JDBC Driver (light)

The concept of being able to query using SQL against Solr seems both crazy and awesome. For some investigations or some administrative tools – it sounds useful though.

Note – there are a lot of new features, and I can’t promise that the list won’t be extended and/or changed. But the above features were the ones that I saw a interesting and can see can be extremely useful.

Stay tuned – and I will share my experiences on the above features in my next posts.

New cool features in Solr that we can use with Sitecore
Tagged on:                                 

Leave a Reply

Your email address will not be published. Required fields are marked *