Amazon server

[kai]

I need to move the APL wiki away from it's current provider towards an Amazon Linux server.

Not a big deal in general but since I played with Amazon servers two years ago the number of server profiles has, well, exploded.

I have a faint idea of how much memory I would need but no idea at all what kind of network performance I would need.

We don't see much traffic on the APL wiki, and stuff that is up/downloaded is almost insignificant by today's standards.

Anybody with experience in this area?

[hebens01]

Hi Kai,

My suggestion is a “try and error” approach to see what works best for the APL wiki.

While you are in the testing phase, it is fairly straightforward to change the “instance type” and you can try different servers. AWS just bills by the hour, so this should not result in extensive costs. Please note though that a change of the instance type requires you to stop and start the server.

I am running a few AWS servers, but most of my traffic is within the AWS network. My experience has been that network performance increases significantly with instances that provide “Dedicated EBS Throughput (Mbps)”. The cheapest instance here is an m4.large, but it may be an overkill and the “t2” instances are perhaps sufficient.

Once you determine the server that works for you, it may be worthwhile to reserve it for a year to save cost.

An item to keep in mind is that disk speed can be a bottle neck. Rather than reserving IOPS, it can be cheaper to get a larger disk (and speed increases with size linearly over a reasonable range).

I would be happy to help further (in which case I suggest you send me a private email). Just trying to give back… there are some APL wiki routines that I use daily.

[kai]

Hello hebens01, I've sent a private message via the forum but have not received a response.

Therefore I wonder whether that did work at all?