Peter Lin wrote:

joking aside, my experience is consistent with Mladen's. my
recommendation as usual is to measure the performance of a given page
and know exactly how long it takes to process the page, get data and
so on.

once you have that, you should be able to predict the performance
fairly accurately.

having said that. it doesn't hurt to have more hardware :)


I agree. From my experience the high-end application that needs to access the database, maintain session, generate pictures and deliver all that to the client takes about 500ms at the current high level CPU (Xeon 3.4GHz).

Interesting is that almost the same response time is constant.
Taking Moore law into equation for hardware, pretty much is the
same with the software. As hardware gets faster, the applications
tend to offer more features.

So if someone wishes to handle a 1000 requests/second it would be
over 3.5 million requests/hour, and that is pretty serious thought.
Also for 1000 request/sec you will be able to deliver only around
8K of data on 100Mbps network, and what would be the purpose of
such a page (not counted requests and headers)?

Regards,
MT.


Attachment: smime.p7s
Description: S/MIME Cryptographic Signature

Reply via email to