mod pagespeed Makes Your Sites Run Up To 2x Faster

[engine]

mod_pagespeed Makes Your Sites Run Faster [googlecode.blogspot.com]

we’re introducing a module for the Apache HTTP Server called mod_pagespeed to perform many speed optimizations automatically. We’re starting with more than 15 on-the-fly optimizations that address various aspects of web performance, including optimizing caching, minimizing client-server round trips and minimizing payload size. We’ve seen mod_pagespeed reduce page load times by up to 50% (an average across a rough sample of sites we tried) -- in other words, essentially speeding up websites by about 2x, and sometimes even faster.

We’re also working with Cotendo to integrate the core engine of mod_pagespeed as part of their Content Delivery Network (CDN) service.

mod_pagespeed integrates as a module for the Apache HTTP Server, and we’ve released it as open-source for Apache for many Linux distributions.

[Alcoholico]

Great stuff provided it does what it promises, my only concern is about installing any thing coming from big bro on my servers. We'd better check sources before installing.

[pontifex]

Very nice - adds all the little things we have done by hand in the past to the apache config. That will help hosters to optimize their systems for the average webspace user!

P!

[CenSin]

I am using shared host, and hope that my provider install it too.

[Pfui]

I agree w/ Alcoholico and have concerns about its functioning. E.g., does it make calls to G's servers like googleapis, etc.? Anyone able to deconstruct a compiled mod?

Also, the mod does a fair amount of page code 'rewriting', something once considered absolutely not okay when browser-based because of the high likelihood it messed up your original code -- alignments changed, ditto font and image sizes, etc.*

mod_pagespeed Filter Examples [modpagespeed.com...]

*Actually, code-changing is still iffy in many quarters because it's still often wrong. Like now, with mobile translators, Or Opera. For example, Opera's "Turbo" feature (emphasis mine) "optimizes pages so they load faster on slow networks. Note that some image quality may be lost when data is compressed. ..." [opera.com...]

[graeme_p]

I see no reason that this would send anything back to Google. Google have never lied about what Chrome does - and you can turn it off.

@Pfui, if you are really worried that the binaries may differ from the source, then just compile it for yourself, or use a Linux distro that compiles it for themselves (none yet, but I doubt it will take long to appear in backports repos). Here is the source [code.google.com ]

This sounds like a very good reason to use Apache (unless the other servers come up with equivalents soon).

[g1smd]

What is the additional server processing overhead needed to run this?

At what point is the trade off, of server load, negatively impacting the site efficiency as a whole?

Anyone run any server load tests on this thing?

[Frank_Rizzo]

Google have never lied about what Chrome does - and you can turn it off

They were economical with the truth with the wi-fi slurp!

[pepperfield]

I'm on a dedicated server and it overloaded our servers and crashed it. Anyone else get this?

[setzer]

If you can, switch to nginx. It's light years ahead of Apache in terms of speed. Stable too, there's even a few top 500 sites that use it.

[Pfui]

@graeme_p: My question was more akin to, say, being curious about the effect of using Analytics, or code from googleapis.com; about whether, at the page code and/or server-mod level, what if any data is sent to G's servers, and why.

I missed seeing the source the first time around (thnx for link), but I'm disinclined to compile (or install) it anyway. Not only because my pages aren't complex enough to benefit, but because that makefile code's usually all G(r)eek to me:)

[tangor]

Haven't investigated it yet, but looks like binaries are offered, not source code... that'll be the biggie for me.

[milosevic]

Does anyone else ever initially react to things like this by thinking along the lines of "Oh no, now everyone will have a fast site, not just people who know what they are doing!"?

:)

[Alcoholico]

I'm on a dedicated server and it overloaded our servers and crashed it. Anyone else get this?

I installed on one server and indeed it behaves erratically, I have not yet pinpointed what is exactly happening but it has to do with large peaks of incoming traffic followed by deep depressions and an abnormal rise in web server clients, the server have not crashed though (yet).

[SevenCubed]

Does anyone else ever initially react to things like this by thinking along the lines of "Oh no, now everyone will have a fast site, not just people who know what they are doing!"?

As a matter of fact that was the exact thought that came to my mind when I saw the thread title yesterday. But since then I've had time to digest it and mull it over and have realized it probably won't have much of a competitive impact for a number of reasons.

First of all most of what that module will do I've already been doing manually. A lot of it is probably based on the Google Page Speed add-on for Firefox which I am a fan of and which I use to help me find ways to tweak a site and/or server performance. But will this mod_pagespeed make it easier for non-technical people to level the playing field? I don't think so because anyone who is aware of the existing Firefox add-on have probably already implemented the suggestions it provides anyway.

Typical sites that I compete against are usually hosted on shared services and often times will not have access to enabling those types of modules if they even knew about it to begin with. Hosting companies will probably be reluctant to enable modules like this just like very few of them had enabled the most basic ones such as mod_gzip/mod_deflate for compression. Why? Because if they have clients on capped resource plans (bandwidth) they loose out on a chance to garner some extra revenue for those who exceed their allotted monthly bandwidth ceiling. Compression modules and/or this new mod_pagespeed carries with it the potential to reduce client bandwidth usage by about 65%. That would be potential surcharges gone up in a puff of smoke. So, I don't think we'll see it applied universally.

In fairness to Google, I appreciate their initiative to try to "speed up the internet" because that ultimately does provide a good end-user experience. All that said, this is another situation where Google is not the creative source of this innovation. I have no doubt that they simple reverse engineered some of the fastest sites on the internet to find out what those techs were doing to accomplish blazing page speed load times then put it all together in this module for mass-distribution -- they didn't have to buy a company for this one though.

[Sgt_Kickaxe]

At face value, thank you Google. It sounds like a promising product.

Reality, I wouldn't let it anywhere near my server without tearing it apart first because you made it and have a history of gathering things without explicit permission. Trust is earned.

[milosevic]

Good points re it being unlikely to become well known, Sevencubed - there's lots of people that don't use Webmaster tools for their site even though it's been around for ages.

[Andem]

I installed it to test it out for a few days. Unfortunately, after an hour, my httpd started hanging even though load was below 1.00. Yeah, I'll wait until it comes out of beta to give it another go.

[gford]

I installed it and my results were horrendous. I can only attribute this to the fact I have highly optimized all my settings already and it is getting confused with all my settings and this module.

My average page speed went up by 4x as much on 10 webpagetest runs.

Just bloody awful..