[Wayne07]

I agree with you that akamai/cloudflare aren't the right kinds of solutions, but once a page of posts fills out (gets to 50), it is static. It doesn't change, doesn't look different depending on who is viewing it. The pages here are great candidates for caching.

You are right that DB clusters and memcache would work. Even simple filesystem caching would probably work. A reverse proxy cache like Varnish would probably take some of the heat off.

[justlurking]

agree with you that akamai/cloudflare aren't the right kinds of solutions, but once a page of posts fills out (gets to 50), it is static. It doesn't change, doesn't look different depending on who is viewing it. The pages here are great candidates for caching.

You can change the length of a page of posts. So, it has to be generated each time, with a query to the database.

Even simple filesystem caching would probably work.

The problem with a simple filesystem would be concurrency. You would have to serialize writes. And, once you create a lot of individual files, you introduce new problems with the metadata.

A reverse proxy cache like Varnish would probably take some of the heat off.

I think FR is already using squid as a reverse proxy, based on some of the errors that I've seen. That may have been in the past.