Google Page Speed Service

Web Performance Optimization Discussions
1

Hi, just enrolled a test Word Press site www.toys-talking.com with Google Page Speed Service, also had W3 Total Cache in operation, and had a few problems.

Just wondering if PSS makes W3 redundant, if not why not?

Cheers

2

No, they are pretty complimentary. W3TC can make your back-end faster which will help PageSpeed Service as well (page, database and object caches). There is a little bit of overlap around front-end resource optimization but a lot less than you’d think.

3

Thanks Pat,

I’ll try it again, I needed to purge the ecommerce section of the website in Google, I was getting pages showing up as incomplete XML pages.

I’ll try it one enabling sections one by one

Cheers

4

I started playing around with Google’s page speed service, and thought the following graphs might be of interest.
I switched over the sites around the 9th and 10th
Here is the link to my AAAAA results
http://www.webpagetest.org/result/130810_QS_V4N/
Although I have almost reached digital Nirvana, the results from Monitis look worse after using PSS
[attachment=312]
This is a Word Press site
[attachment=311]
This is just a small site produced with Bootstrap
Both sites are hosted on our own testing server in our office, running on a DSL line

5

As best as I can tell, Montis is only monitoring the base HTML performance, not the full page performance. Do you have any real-user monitoring on the pages (Google Analytics, mPulse or something like that)? That should give you a much better idea of the end-user performance.

Any accelerator is usually going to make the base page a little slower because it has to go through the accelerator and back to the origin instead of directly to the origin. Most of them can easily make up the additional delay and more with the optimizations they do to the page and in the serving of the front-end resources.

That said, the base page can also sometimes get faster if the content is cacheable on the edge. There are also various PSS filters that can cache part of the HTML and serve it while still waiting for the back-end response from the origin (so TTFB will be much faster and external content will start downloading even if the base HTML overall takes longer).

Something like Montis is good for measuring availablity and consistency of the server response times but isn’t terribly useful for understanding the user experience.

6

Thanks Pat,

I’ll load up another site, that has had Google Analytics running on it for some time.

Wait a few days and see what the results show.

One thing I do like about PSS is the time saving as well, I don’t need to upload my image files to Gogrid, compress the css, js and upload that as well, quite a time saver!