Thanks for starting this thread Nicolas. I think it’ll be very useful for other projects.

I am currently helping with some of the work for the Atlas of Living Scotland (AoLS) which we hope to have running in September. This will be ran on EC2 and Im putting together the deployment. I plan to document/blog about this deployment so others can use it for reference with the used specs of the EC2 machines etc.

To address the points:

* Were those requirements sufficient for you? These specs are only suitable to get a demo up and running and are not sufficient for production.

* Did you deploy everything on a single server or did you use a more splitted infrastructure? If so, did you wrote your own ansible scripts to match your configuration? We do not recommend everything is installed on a single server for production. Specifically, SOLR and Cassandra both benefit from running on their own dedicated machine. The work for AoLS has made me realise that while the demo ansible playbook [1] is useful for folks doing a quick demonstrator, it isn’t very helpful as a starting point for starting production deployments. The ALA developers use separate playbooks for each component [2] and we need some documentation around these playbooks. I think we need a "Getting Started” for production deployments which would give a list of unusable playbooks to run.

* Did you use physical servers? Or Amazon EC2 instances ? Or another cloud providers? I think EC2/Digital Ocean/other-colour-provider instances should be fine.

* Is there anything you'd do differently if you had to do it again tomorrow? Any other comments or remark? I guess I'll leave this for my blog/documentation.

Thanks again,

Dave

[1] https://github.com/AtlasOfLivingAustralia/ala-install/blob/master/ansible/al... [2] https://github.com/AtlasOfLivingAustralia/ala-install/tree/master/ansible

On 19 Aug 2015, at 10:26 am, Nicolas Noé <n.noe@biodiversity.bemailto:n.noe@biodiversity.be> wrote:

Hi guys,

I'm still evaluating how to deploy an ALA instance for Belgium at the end of this year and was interested in hearing your first hands experience about the infrastructure you run the portal on.

I'm especially interested about the experience of countries like France and Spain that did it recently and who have a volume of data of the same order of magnitude as Belgium.

The documentation (https://github.com/AtlasOfLivingAustralia/documentation/wiki/Essential-tools...) recommends the following server configuration:

- 100Gb SSD - 32GB RAM - 2x CPU

My questions are:

* Were those requirements sufficient for you? * Did you deploy everything on a single server or did you use a more splitted infrastructure? If so, did you wrote your own ansible scripts to match your configuration? * Did you use physical servers? Or Amazon EC2 instances ? Or another cloud providers? * Is there anything you'd do differently if you had to do it again tomorrow? * Any other comments or remark?

Thanks a lot for your collaboration!

Nicolas

_______________________________________________ Ala-portal mailing list Ala-portal@lists.gbif.orgmailto:Ala-portal@lists.gbif.org http://lists.gbif.org/mailman/listinfo/ala-portal