Is the TTFB for different locations and browsers irrelevant if you are self-hosting?

[danatanseo]

Please forgive my ignorance on this subject. I have little to no experience with the technical aspects of setting up and running a server.

Here is the scenario:

We are self-hosted on an Apache server. I have been on the warpath to improve page load speed since the beginning of the year. I have been on this warpath not so much for SEO, but for conversion rate optimization. I recently read the Moz Post "How Website Speed Actually Impacts Search Rankings" and was fascinated by the research regarding TTFB. I forwarded the post to my CEO, who promptly sent me back a contradictory post from Cloudflare on the same topic. Ily Grigorik published a post in Google+ that called Cloudflare's experiment "silly" and said that "TTFB absolutely does matter."

I proceeded to begin gathering information on our site's TTFB using data provided by http://webpagetest.org.  I documented TTFB for every location and browser in an effort to show that we needed to improve. When I presented this info to my CEO (I am in-house) and IT Director, that both shook their heads and completely dismissed the data and said it was irrelevant because it was measuring something we couldn't control.

Ignorant as I am, it seems that Ilya Grigorik, Google's own Web Dev Advocate says it absolutely is something that can be controlled, or at least optimized if you know what you are doing.

Can any of you super smart Mozzers help me put the words together to express that TTFB from different locations and for different browsers is something worth paying attention to? Or, perhaps they are right, and it's information I should ignore?

Thanks in advance for any and all suggestions!

Dana

[vmialik]

Hi Dana,

Yes TTFB is something you can control with the type of server you use. And where that server is in relation to your visitors. You cannot control the browsers they use, but hear are some thoughts on possible optimizations:

Server side: http://createdevelop.blog.com/2010/10/12/how-to-reduce-time-to-first-byte/

Location (plus other suggestions): http://msdn.microsoft.com/en-us/magazine/dd188562.aspx

[danatanseo]

Thanks Vadim. This is helpful. In the first article the author writes:

"The only thing that is controllable is the server you are on." He suggests optimizing the database. What specific & measurable directive might I give to our IT manager that would accomplish this goal?

The second post looks very helpful indeed. I am downloading Microsoft's VRTA right now. It's a bit technically over my head, but I get the concepts. This should be something I can pass on to IT...however, it seems the info could be a bit dated (it repeatedly references IE 7)...Is there anything additional that might be more current?

Thanks again!

[vmialik]

Databases?  Optimize any database queries that are slow This should help: http://www.techfounder.net/2011/03/25/database-profiling-and-optimizing-your-database-the-generic-version/

Now before you pass anything over to the IT this issue is a heated one in some cases where you have people saying that TTFB is not might not be the key metric to go after, here is more food for thought:

http://blog.cloudflare.com/ttfb-time-to-first-byte-considered-meaningles

"At CloudFlare we make extensive use of nginx and while investigating TTFB came across a significant difference in TTFB from nginx when compression is or is not used. Gzip compression of web pages greatly reduces the time it takes a web page to download, but the compression itself has a cost. That cost causes TTFB to be greater even though the complete download is quicker."

[danatanseo]

Thanks Vadim. Yes, this Cloudflare post is exactly the one I was referencing in my question. As I mentioned, Ilya Gregorik posted a rebuttal to their experiment here: post in Google+

It seems to me that if a Google developer says TTFB absolutely does matter that this would take precedence over anything Cloudflare might say.

What do you think?

[vmialik]

Yea this makes sense as others have said that Cloudflare is trying to say that TTFB is not the most important metric, and so they published this study, as it aids their business model.

I would do just that listen to Google dev vs Cloudflare. Also the way I think about it even if their studies are true, where for the overall benefit TTFB would have to increase if you are using some compression, you still need to work and decrease your TTFB either way, that is just intuition. I apologize if I made it seem that TTFB is to be ignored, because Cloudflare state's that quite boldly,

Again some things that affect TTFB:

1. Move your website to a faster/better server (If an option)
2. Use a CDN or something similar to reduce the load on the server (repeated requests to a server will increase the TTFB)
3. Reduce the time the server spends processing the request for information (sent above) and more here: http://stackoverflow.com/questions/10938682/how-to-reduce-server-wait-time

[Igal_Zeifman]

I absolutely agree with Vadim. (+1)

Google is the best source for Google facts. Everything else is just speculation.

And yes, generally speaking, the best answer is to use a CDN.... 
The reason is simple. CNDs proxy technology, which was designed to minimize "physical" distances between the site's content and browsers, directly influences TTFB.

Being an in-house SEO for a CDN company I get a lot of questions about this from our support and clients. I have to admit, until recent Moz post, I wasn't aware of full implications of TTFB and considered it to be one of few page load speed related metrics.  (http://moz.com/blog/how-website-speed-actually-impacts-search-ranking)

This post really helped me get a better grasp on things. Interestingly enough, few month ago one of our clients Guest Posted in our blog about speed improvement gained by our free plan. Among other things, he mentioned 70% improvement in TTFB (grade going from F to A) 
(http://www.incapsula.com/the-incapsula-blog/item/718-what-incapsula-free-did-for-my-site)

At the time I didn't give it much attention. Because, like many others, I was focusing on overall load speeds.... 
Now I can't help but feel that this was a missed opportunity.
This post could be even better with the added SEO angle...
If anyone here is interested in giving this a try and guest posting about it, I`ll be happy to provide all resources needed on our end.

[vmialik]

Hi Igal,

Do you work for incapsula, you are mentioned as a Mod on the blog?

I have heard great things about incapsula from others, but in terms of TTFB is it better than cloudflare? If so, how so?

Also any other ways that it excels Cloudflare? any ways its inferior to Cloudflare in your opinion?

Thanks I am really looking for more info, as I had great results with Cloudflares features and offering, wondering if I should give Incapsula a run

Thanks

[Igal_Zeifman]

Hi Vadim

Thanks.
Yep, I work for Incapsula but no, we are not the said "Mod".  

As for CF comparison... Generally speaking, we are more business oriented and security focused. I know that our security offering is more comprehensive, especially because both WAFs were comparatively pen-tested on several occasions and we always came out as consistently (and significantly) better option.We also have addition security features - like 2FA support and backdoor shell protection - which CF simply doesn't offer and we do more in way of ddos mitigation, especially against smart application layer attacks which require security capabilities, besides network muscle.

Still, speed wise, I always considered us to be pretty much on the same level. However, until few days ago I never considered TTFB to be such core SEO factor, so maybe we have better performance there...  
But again, to be fair, I`m only speculating - mostly based on the CF blog you've shared.
(if TTFB is considered un-important, it might also be under developed...)

Might be an interesting thing to test and document.

[vmialik]

security wise it seems both of you guys have stellar options. for me the issue is performance, caching for dynamic sites, CDN performance, and in this case TTFB response.  I was not sure with your response do you have faster TTFB to CF?

Thanks

[Igal_Zeifman]

Honestly, I don't know. I don't think TTFB was ever comparatively tested - at least no to the best of my knowledge.

For security, these are some of the resources I can point to.
I understand that this is not the main issue
Still, I wanted to provide some factual context to my previous statements.
http://zeroscience.mk/files/wafreport2013.pdf http://ddos-protection-services-review.toptenreviews.com/ http://tonyonsecurity.com/2012/11/13/protecting-your-website-cloudflare-or-incapsula/ 
(This last one is interesting since Tony is a COO of Sucuri. Some would call his our competitor. I prefer 'colleague' )

[danatanseo]

Many thanks to both Vadim and Igal for such great information and also a really great thread on the subject. I really, really appreciate your answers.!

[vmialik]

Thanks Igal and hopefully you have some info to work with Dana!

[danatanseo]

Yes, very helpful guys. I appreciate it!