TTFB grade - varies within webpagetest.org sites

[ryankent]

Hello,

I setup a clean WP install and added 2 extensions, then tested the site speed. We are working with our developers and server techs to consistently achieve page loads under 2 seconds. We will add a CDN later, but first we wish to optimize the site as fully as possible.

Test Result 1 - http://www.webpagetest.org/result/131027_SJ_JQF/
San Jose using FF

Shows TTFB as 0.342 seconds with an "A" grade. The second test shows TTFB as 0.738

Test Result 2 - http://www.webpagetest.org/result/131027_GP_JR6/
Asheville, NC using IE 8

Shows TTFB 0.313 (faster than the first test) with a "C" grade. The second test shows TTFB as 0.186 which is again faster than the first test.

Q1 - Why does the Asheville site show a "C" letter grade for a speed faster than the San Jose server does with a slower speed?

Q2 - This is a clean WP install with 2 extensions:
http://wordpress.org/plugins/rpx/
http://wordpress.org/plugins/w3-total-cache/

The site is on a dedicated server based in Orlando, FL. The goal for this site is to achieve a perfect page sped score using Google Page Speed. Why? We are trying to push the envelope in terms of optimizing page speed and this test site is being used to gain that understanding.

The only issues showing are below. My question is whether others feel resolving those issues are likely to reduce page load times. We do not wish to blindly follow a tool's recommendations. If you do feel it would reduce page load times, how can we make these adjustments? I presume if they could easily be made, WordPress would have done so within their install package.

Thank you,

Remove render-blocking J

http://ryankent.co/wp-content/cache/mini...RaGW9AQ.js
Optimize CSS Delivery of the following:

http://ryankent.co/wp-content/cache/mini...wkB6kX.css
http://fonts.googleapis.com/css?family=S...Clatin-ext

[pmeenan]

The TTFB target uses the socket connect time as a baseline for the RTT to the server to know how much time to allow for the request round trip. Ashville must be closer to the server than San Jose (or something caused the San Jose connect to take longer).

[ryankent]

Thank you for the response Patrick.

The server is located in Orlando, Florida so yes, the Ashville server is closer. Even so, why would a slower TTFB time receive a better grade than a faster time?

[robzilla]

Quote:Even so, why would a slower TTFB time receive a better grade than a faster time?

Probably because the TTFB is farther removed from the Target First Byte Time.

342 ms First Byte Time
291 ms Target First Byte Time

vs.

313 ms First Byte Time
84 ms Target First Byte Time

The numbers vary quite a bit when you rerun the test.