>I can't comment if there is a performance difference between the open
>source version and professional version. If there is a difference, I
>doubt you would see it in testing. Maybe someone else can comment on this.
It depends on how your app is structured. If you're using lots of internal
HTTP requests, you can make them "conditional" HTTP requests. If you're
using Resin Pro on the target of those conditional GET requests, and your app
knows how to invalidate the cache only when the response data is changed, you
can see some real benefit. If there has been no change, Resin Pro will
respond with a 304 - Not Modified. In other words, the majority of request/
response cycles would be comprised only of HTTP headers with no message body.
The open-source version, without some intermediary like Squid running in
server-accelerator mode, can only respond to the request by executing
whatever process is involved to generate the response, and sends HTTP
responses which always include the message body. By using Resin Pro to cache
the output from your process, you gain two advantages over Squid: First, the
response is cached as an output stream, not written as a file. Second, Resin
Pro's cache is threaded, Squid is not.
For the app we're developing using a REST approach, Resin Pro should perform
better in testing, since without it the application would not scale well
under load as it's basically a recursive HTTP request using chained filters.
The tradeoff is a larger RAM requirement, which I'll take if it saves a lot
of HD reading and writing or CPU cycles.
resin-interest mailing list