SLOW Demo Web Speed Statistics and Admin Display Page Parse Time BlueHost

[Zean]

I am trying to figure out why my Zen Carts are so slow. As you may see from the chart, it took an average of 3810ms to show the default home page for the default demo data installation of Zen Cart v154 on my BlueHost Shared Plus hosting account. This was measured from 10 monitoring severs (www.freesitestatus.com) around the world. As shown in the chart almost all the time is due to "PROCESSING" (see below for definition).

However, when I display the Page Parse Time (https://www.zen-cart.com/content.php?145), here are some examples of what I get . . .

Parse Time: 0.540 - Number of Queries: 495 - Query Time: 0.32343136608887
Parse Time: 0.479 - Number of Queries: 481 - Query Time: 0.29987357475281
Parse Time: 0.346 - Number of Queries: 502 - Query Time: 0.15993407011414
Parse Time: 0.351 - Number of Queries: 484 - Query Time: 0.16587302929687
Parse Time: 0.767 - Number of Queries: 462 - Query Time: 0.25757795991516

How can I use this to understand why the Zen Carts are so slow and what I might specifically ask from my web host?


This is the default demo data installation of Zen Cart v154.
Host: BlueHost "Shared Plus" account
Zen Cart v1.5.4
No plugins
Zen Cart installed via FTP and run zc_install
PHP Version: 5.4.43 (Zend: 2.4.0) PHP Memory Limit: 128M
Database: MySQL 5.5.42-37.1-log



Definitions of the performance parameters measured in the chart . . .

REDIRECT
The time required to follow all redirections (if any) before the final URL.

DNS
The time required to resolve a Fully Qualified Domain Name (FQDN) into an IP address. If you stated an IP address as the destination, then this value will be 0.

CONNECT
The time required to establish a TCP connection.

REQUEST
The time required to send a request.

PROCESSING
The duration from sending the request to receiving the first byte of response. This gives an indication of the processing time or responsiveness of the website/server/ service.

DOWNLOAD
The time required to download the entire response from the first byte till the last byte. This gives an indication of the bandwidth available.

[balihr]

Those are actually pretty good results, if you ask me. You're on one of the top 10 WORST hosts and you're on a shared account, sharing CPU power and RAM with thousands of other sites... I'm not really sure what kind of results you're expecting, but you should keep in mind that you get what you pay for. Your plan is what, $5 per month? It's like being surprised after a Ferrari overtakes you, and you just bought a 15 year old Hyundai from a used car dealer...

Here are my stats on an entry level VPS (which costs me $10 per month, BTW):
Attachment 15592

The processing time indicates time required for server to compile the code and generate the HTML output which is sent to your browser after the initial response. The more "power under the hood" your server has, the faster it will be in compiling the code. Your server is heavily loaded with A LOT of other accounts and is running A LOT of websites at the same time so you don't get that much juice... Seriously, 3.8s on Bluehost is not a bad result...

[Zean]

Thanks! It is interesting to see your Zen Cart store's web performance. Could you also post your Page Parse Time?
Admin->Configuration->Logging->Display The Page Parse Time

1. BlueHost is recommended by the owner of the Zen Cart forum for hosting Zen Cart stores . . . https://www.zen-cart.com/content.php?3-services
2. I pay $144/year in advance which is $12/month just for hosting . . . and is more than you pay . . . plus, additional amounts for dedicated ip, etc.
3. 3,810ms to show a home web page is poor performance and will cause customer experience to suffer and a google ranking penalty.

However, I would like to focus on the technical questions. So getting back to my original question . . .

4. I have done a bit of experimentation. Outside of the Zen Cart store, I created a static web page version of my Zen Cart store home page. So, it does not access the database but presents the same screen and images, etc to the user. It's web performance is much faster 679ms using the same 10 monitoring severs (www.freesitestatus.com). How can I use this performance and parse data to understand why the Zen Carts on my server are so slow and what I might specifically ask from my web host?

[lhungil]

I am trying to figure out why my Zen Carts are so slow. As you may see from the chart, it took an average of 3810ms to show the default home page for the default demo data installation of Zen Cart v154 ...

The parse and database times are reasonable from what I've seen for the average shared hosting plan. Almost 2.5 to 3 seconds added after PHP has finished processing seems a bit high to me... Makes me wonder if the hosting provider has some network issues or bandwidth limits in place... Or possibly some sort of communication issue between the Web sServer and PHP... Or possibly some sort of proxy / cache involved...

Out of curiosity what does webpagetest.org show? I find it tends to be useful for performing page speed tests (especially since one can set the "client" download connection speed and the results are broken down by first-byte, start of browser render, visually complete, and fully loaded).

I also like how webpagetest.org shows you performance based on new vs repeat views of the page. The results will also include an optimization score and a fairly detailed breakdown of ways to possibly improve the site performance.

Disclaimer: I am not a paid advocate even if the above sounds like a sales pitch. I should also note the above site is just one of many I use to analyze website performance. I should also note the webmaster tools built into most modern browsers can also be very useful for determining speed based upon your current computer, browser, location, and bandwidth.

... Here are my stats on an entry level VPS (which costs me $10 per month, BTW):
...

O.o, not a bad price for an entry level VPS... Is the VPS hosting provider managed or self managed?

I (and a few clients) are paying around the same for shared hosting (during a sale runs as low as $6.99 a month + fee for dedicated IP). As expected the performance tends to vary based on the workload of the server. During peak US times the Zen Cart sites average just around a second for a complete page load (worst I have seen so far was around 2 seconds). During off-peak US times, complete page loads are often around half a second or less (screenshot below taken around 8:30PM CDT).

[balihr]

Thanks! It is interesting to see your Zen Cart store's web performance. Could you also post your Page Parse Time?
Admin->Configuration->Logging->Display The Page Parse Time

Parse Time: 0.336 - Number of Queries: 476 - Query Time: 0.12452050010681
Parse Time: 0.317 - Number of Queries: 497 - Query Time: 0.13381776307678
Parse Time: 0.336 - Number of Queries: 462 - Query Time: 0.13807176324463

1. BlueHost is recommended by the owner of the Zen Cart forum for hosting Zen Cart stores . . . https://www.zen-cart.com/content.php?3-services

You need to understand it's an affiliate link and since very few people donate to the project, the authors need to get some cash from other sources if for nothing else, then to pay the bills to keep the project active (I'm pretty this forum runs on a dedicated server...)

2. I pay $144/year in advance which is $12/month just for hosting . . . and is more than you pay . . . plus, additional amounts for dedicated ip, etc.

Some of my clients pay 80 GBP (which is around $120) per month for their Cloud hosting. But, there are no performance issues. It's irrelevant if you're paying $5 or $10 or even $15 per month, it's still cheap and you get what you pay for...

4. I have done a bit of experimentation. Outside of the Zen Cart store, I created a static web page version of my Zen Cart store home page. So, it does not access the database but presents the same screen and images, etc to the user. It's web performance is much faster 679ms using the same 10 monitoring severs (www.freesitestatus.com). How can I use this performance and parse data to understand why the Zen Carts on my server are so slow and what I might specifically ask from my web host?

You can't compare serving static pages with dynamically built content. Static pages are already there and are just waiting to be served. Dynamic sites (such as Zen Cart) need to be composed from various elements and all merged into a static page before it's served to the browser...

O.o, not a bad price for an entry level VPS... Is the VPS hosting provider managed or self managed?

It's self managed and using Plesk. It auto-updates, has an automated backup on external drives... In the 3 years that I've been there, the only management that I needed was when I was playing with some Apache mods and installing mod_pagespeed and tried to install inotify (and failed with the latter). Other than that, it just works... Very little downtime - according to Pingdom, I'm at 99.94% uptime.

I (and a few clients) are paying around the same for shared hosting (during a sale runs as low as $6.99 a month + fee for dedicated IP). As expected the performance tends to vary based on the workload of the server. During peak US times the Zen Cart sites average just around a second for a complete page load (worst I have seen so far was around 2 seconds). During off-peak US times, complete page loads are often around half a second or less (screenshot below taken around 8:30PM CDT).

I actually lied. It's 10 EUR, so $11. It comes with a dedicated IP, 2 cores, 2 GB RAM (burstable to 4), 250 GB disk (or 125 GB SSD) and 500 Mbit port speed. But, self managed and located in Germany. I'm really happy with it, but I wonder how much I'd gain in speed if I moved to their SSD (and it's only a ticket away, but I'm just too lazy).
However, looking at your screenshot, it makes me scratch my head... HOW?! I wonder what would happen if I turned on mod_pagespeed on that domain, but that's a different story now...

[Zean]

The parse and database times are reasonable from what I've seen for the average shared hosting plan. Almost 2.5 to 3 seconds added after PHP has finished processing seems a bit high to me... Makes me wonder if the hosting provider has some network issues or bandwidth limits in place... Or possibly some sort of communication issue between the Web sServer and PHP... Or possibly some sort of proxy / cache involved...

Thanks! I'm on a bit of a learning curve . . . How is it that you are able to infer "2.5 to 3 seconds added after PHP has finished processing?"

Out of curiosity what does webpagetest.org show? I find it tends to be useful for performing page speed tests (especially since one can set the "client" download connection speed and the results are broken down by first-byte, start of browser render, visually complete, and fully loaded).

I ran webpagetest.org as suggested . . . Their performance grading of the page was . . .

Grade
F - First Byte Time
A - Keep-alive Enabled
A - Compress Transfer
N/A - Compress Images
A - Cache static content
X - Effective use of CD

Thoughts?

Parse Time: 0.336 - Number of Queries: 476 - Query Time: 0.12452050010681
Parse Time: 0.317 - Number of Queries: 497 - Query Time: 0.13381776307678
Parse Time: 0.336 - Number of Queries: 462 - Query Time: 0.13807176324463

Thanks for posting your Page Parse Time and query times so I can compare with mine. I think yours are fairly similar to mine even though we may be comparing apples to oranges since mine is from a default installation of the zen cart demo store data (I used the default demo store so there could be a consistent basis for comparison since anyone can install the same demo data on their server) and I'm not sure how that store compares to your store. In any case they are fairly similar. So the overall performance difference is due to something other than those "Parse Time" and "Query Time"

You can't compare serving static pages with dynamically built content. Static pages are already there and are just waiting to be served. Dynamic sites (such as Zen Cart) need to be composed from various elements and all merged into a static page before it's served to the browser...

Actually, I can compare "serving static pages with dynamically built content" so I can see what additional time is associated with the dynamic page above and beyond serving the same page as a static page. I thought this might be helpful as a first step in isolating the source of the bottleneck.

[lhungil]

From the article "What are Parse Time Numbers":

The page Parse Time is the time it took from when the first line of code in index.php started processing until the footer of your page was drawn. This includes the delays introduced by any SQL queries waiting on the database. It is the sum of both Query Time and all the time it takes to request data from the database and process the information it receives back. So CPU time is the difference between Parse Time and Query Time.

Basically the Parse time is the amount of time it takes PHP to do 98% of the DB calls and HTML output for the page (except some session cleanup, garbage collection, and destructors - minor at best) on a stock Zen Cart (by default the end time is calculated when outputting the footer - very close to the end of the PHP script). So the question is "Why does the request take 2.9 - 3.6 seconds when PHP is mostly done processing and outputting the HTML after .35 - .46 seconds?".

Could speculate for hours on other items which could be responsible for the delay... Might just need a tweak of the server configuration (settings for PHP, Web Server, Web Server <=> PHP interface, etc)... Or could be much more complicated... Probably best to ask the Hosting Provider as they hold the keys to the kingdom (including access to logs, server configuration, network configuration, etc) and are thus in the best position to answer the question and optimize as neccessary. Unless of course one prefers to just jump ship and try another Hosting Provider.