As an existing hardware client, you can buy one-year license renewals throughout 2017. However, if fancying a change, here's 4 important steps for switching to Apache Solr or Elasticsearch.
1. Choose your hosting solution and servers
Whereas Google Search Appliance comes ready to plug in, Apache Solr and Elasticsearch need to be deployed and hosted on servers, on premise or in the cloud. Both platforms are highly scalable and can be massively distributed.
Open source on premise
Servers and hardware requirements are highly dependent on your number of documents, document types, use cases and number of users. Memory, CPUs, disk and network are the main parameters to consider. Both Elasticsearch and Apache Solr require running java.
Open source in the cloud
You can also choose to run Apache Solr or Elasticsearch on a cloud platform.
For SolrCloud, you need to install Zookeeper.
2. Define your schema and mapping
In Apache Solr and Elasticsearch, fields can be indexed and processed differently depending on type, language, use case ... A field and its type can be defined in Elasticsearch using a mapping API and in Apache Solr with the schema.xml.
3. Tune your connectors
Do you need to change all connectors?
The answer is no. Connectors sending GSA feeds can be kept, just refactor the output to match the Elasticsearch or Apache Solr indexing syntax.
However, if you use GSA to crawl websites, you need either to reconsider crawling as a method or to use an external webcrawler (Norconex). Contrary to GSA, Apache Solr and Elasticsearch do not come with a webcrawler.
4. Rewrite your queries and fetch new output
All common query functions such as filtering, sorting and dynamic navigation are standard in both Apache Solr and Elasticsearch. However, parameters and output (XML or JSON) are different, which means queries and front-end need adaption to your new search engine.
If you are using Jellyfish by Findwise, queries and output will roughly be the same and adaptation only means mirroring what's already been done.