Bearing in mind that the ElasticSearch-Zookeeper plugin doesn't support v0.90 release. With unicast, what's your strategy on updating your list of IPs? I.e. when upgrading/scaling-up/down.
What client-side connectivity (from web/worker role) to the cluster? Do you:
a) implement your own round-robin/failover implementation across all nodes in the cluster
b) spin up a local (non-data/non-master) elasticsearch process on the client machine that joins the unicast cluster. The application will only connect to localhost
Where do you store your data? Azure blob gateway?
Can you share your detailed story on your ElasticSearch experience on azure, and any particular points/issues to watch out for?
Cheers
We recently announced that users can find, deploy, and manage Elasticsearch from within the Azure portal.
The cloud platform where you'll deploy your deployment. We support: Amazon Web Services (AWS), Google Cloud Platform (GCP), and Microsoft Azure. You do not need to provide your own keys. The cloud platform's region your deployment will live.
Next stepsCreate an Ubuntu VM in an Azure resource group. Install Elasticsearch, Logstash, and Kibana on the VM. Send sample data to Elasticsearch from Logstash. Open ports and work with data in the Kibana console.
Just a note about this. We are on the way on releasing Azure plugin for Elasticsearch. It will help to allow automatic discovery of your Elasticsearch nodes. I think that we will have something public in the next weeks.
Also, I recommend to use local storage. Azure blob will be used in the future to allow snapshots (and restore) feature when Elasticsearch 1.0 will be out.
Hope this helps
Update: Plugin is now available here: https://github.com/elasticsearch/elasticsearch-cloud-azure
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With