Path: csiph.com!fu-berlin.de!uni-berlin.de!individual.net!not-for-mail From: Arno Welzel Newsgroups: comp.lang.php Subject: Re: PHP boilerplate to appear at the top of any program Date: Tue, 5 Jan 2016 18:55:55 +0100 Lines: 41 Message-ID: <568C03AB.10908@arnowelzel.de> References: <568AE9F9.5010506@arnowelzel.de> <568B365C.6080100@arnowelzel.de> <87bn90cxw5.fsf@ahungry.com> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit X-Trace: individual.net ttIQjYxLpeMlK1zMnhTwrw/053HvpKctNmYY/4Az//eZb9aIpU Cancel-Lock: sha1:brzutMzNlGLoTZNoe2u6kjvq1YQ= In-Reply-To: Xref: csiph.com comp.lang.php:16090 Jerry Stuckle schrieb am 2016-01-05 um 15:08: > On 1/5/2016 2:02 AM, Matthew Carter wrote: [...] >> Also, we are discussing PHP - as javascript assets are not even >> necessarily from the same server that is running the PHP script (ever >> hear of a CDN?) they are not a factor, as they only impact client render >> speed - if you pulled in a site via a curl request, or a non-graphical >> browser such as lynx, you wouldn't need to request all those assets either. >> > > Yes, I've heard of CDN. But Arno isn't using it, as aren't others. And Oh - you even *heard* of CDN. Wow. JFTR: Our customers are *using* CDNs (and cloud services for stress testing as well to simulate 10.000 clients accessing a server at the same time - but maybe this is also something you just heard of and never used yourself). > the various frameworks which have been discussed don't use it. So your > comment is immaterial. And if you pull a site via curl or lynx, you're > not doing page views. The test was about how much PHP scripts can be handled per second - nothing else. And of course the number of page views is NOT higher than that and the test is in no way useless as the real world the number of possible page views will NEVER be higher than this. So I still say, that using Apache Bench gives good impression for what will happen when changing things in the examined PHP scripts. Just introducing a database connection which has to be established with every request may dramatically slow things down - and Apache Bench will clearly show this. Sometimes even the overhead of a framework alone is not as severe as just connecting to a database for every request and not re-using persistent connections. -- Arno Welzel http://arnowelzel.de http://de-rec-fahrrad.de http://fahrradzukunft.de