Specs for stable public server

[b]

Assess what Amazon configurations can be used to support the DJANGO infrastructure as it stands (e.g. without additional server-side optimization such as load balancing, different database technology etc.).

Generate some benchmarks for single-user query performance on several default instance types: e.g. micro, small, medium, large.

Use data to decide if architectural changes are needed and what stable server set up to use.

[Richard Bruskiewich]

Benchmarking of various AWS EC2 suggests that we can reduce from the 122GB server down to a 30GB server with acceptable performance at current prototype level of usage (no multi-user stress testing yet...). This is:

r3.xlarge 4 13 30.5 1 x 80 SSD $0.350 per Hour which costs about US 260 per month (for an demand instance), NOT including EBS storage costs

It is tempting to eventually split the web server from the database, reducing the size of the web server instance (since the Django app is not too complex) but keeping sizeable RAM in the back end database server.

[Richard Bruskiewich]

I think we have a tentative decision to use an r3.xlarge 30.5 GB server for now. This issue can be reopened again later, when we get more experience with the performance of the system.

[Richard Bruskiewich]

I think the current 30gb instance is working well. The only outstanding aspect of this decision is to perhaps move the server instance to Scripps AWS