Our site is slow..
-
We have noticed that our site is much slower than many of our competitors and expect that this is affecting our organic ranking.
We are on dedicated UK Fast server with
[20/08/2013 16:44:35] Tony Jackson: Operating System: CentOS 6 64-bit
CPU: 1x Intel Xeon E5-2620 @ 2.0GHz
Memory: 8x 8GB DDR3 RDIMM
Hard Drive: 4x 128GB SSD RAID10
[20/08/2013 16:46:28] Tony Jackson: Nginx, percona, tomcat/solr, Magento CE, cat: 25200 productsI could really do with your thoughts on this page for example
http://www.pretavoir.co.uk/sunglasses/ray-ban-sunglasses.html
-
I am not a huge fan of APC at present (PHP PECL library, not the power battery company). Maybe nginx won't segfault like Apache has been for us.
As for the hosting, what I mean is to use a load balancer with multiple servers. Will you take some performance hit in getting DB data? Probably, but very little. But you're getting far worse now with Magento so it doesn't matter how fast your machine is (like I said, it's a beast so if it's not sitting at 0.01 almost all the time you've got major issues). Remember that this impressive machine is also a single point of failure. I had an impressive box running our DB once (multiple webservers) and the RAID controller crapped out in the middle of the day. We got back up under the two hour SLA but it cost us a lot of sales.
Now we run two webservers, a multi-AZ DB (automatic failover to secondary if master dies) and all behind a load balancer in AWS. I won't pretend this isn't without any issues (we wound up in an odd instance where the DB failover didn't work as expected) but it's a lot more redundant and scalable. I bet our setup is cheaper than your behemoth, too. Remember, you still have to chew through all the code each and every time the page is loaded (your problem is processing the individual thread, not horsepower). Splitting your site amongst multiple webservers means you have scalability as well.
-
Highland, Thank you for the response.
We are using memcached locally as it seems to give best results, having tried a combination of APC and memcached.
Would moving our database to a different server speed up the time to first byte(ttfp) as currently we sit with very little load on cpu? What would be the benefit?
Where you mention 'a cloud based solution' are you referring to using Cloudflare to sit in front of our server to act as a cdn and cache, or to move to a complete cloud based solution? I assume it is their Business package you refer to which gives the ddos and other protection.
Looking at Googles pagespeed tools, out ttfp seems to be the hold-up. which then just leaves the off site code (Snapenage, Comodo, etc) bringing up the last elements.
-
Thanks for this response.
We had been looking at Cloudflare and also at rearranging our server somewhat. Will let you know.
-
Magento is built on top of Zend Framework 1, a notoriously slow PHP framework. It also does some things that slow it down. Some thoughts...
- Be sure your use the Magento caching system.
- Move your database off to another machine. 25k products is a LOT
- Your hosting setup is monolithic. While your box is a beast of a machine (12 threads, 64GB RAM, 4x 128GB SSD) it's still susceptible to slow-down. I bet 50 concurrent users could bring it to its knees. While CDNs can help offload some of this, I recommend the beast with a cloud based solution. I bet you could get 2 or 3 servers plus a DB server for less than what you're paying for this one beast. Then put your machines behind a load balancer. So you would have multiple web servers serving your site up, a load balancer routing traffic to the fastest one and a common DB powering it.
- Switch to Cloudflare for your DNS. They will cache pages for you and they do it for free. They also add a layer of security.
- Use WMT to benchmark your site speed. It tells you the load time in a handy graph. Play around with it and see what happens.
-
We are using Magento on Nginxwth caching enabled with memcache
-
Generating the page seems to take very long. What shopping cart software are you using? Can it cache pages?
-
Thanks for your reply.
This page is a better example of our problem
http://www.pretavoir.co.uk/sunglasses/ray-ban-sunglasses.html
-
Your website isn't that slow from London but there lots of things you can do to improve speed:
1. Use a CDN like NetDNA to host CSS, Javascript and images. By using a CDN, you will also increase the number of simultaneous (parallel) downloads since these files will be hosted on another (sub)domain.
2. You have lots of Javascript files. Try combining them into one file.
3. Add a cache header, this will instruct web browsers to cache your images, css and Javascript files. Simply add this to your .htaccess file:
AddType image/x-icon .ico
<ifmodule mod_headers.c=""># YEAR
<filesmatch ".(ico|gif|jpg|jpeg|png|flv|pdf)$"="">Header set Cache-Control "max-age=29030400"</filesmatch>WEEK
<filesmatch ".(js|css|swf)$"="">Header set Cache-Control "max-age=604800"</filesmatch>
24 HOURS
<filesmatch ".(html|htm|txt|php)$"="">Header set Cache-Control "max-age=86400"</filesmatch></ifmodule>
======================================================
... and sorry for my english.
Hope that helps!
-
Yes had looked at GT Metrix though not paid much attention to YSlow. Will take a look now.
However it may be of more interest to look at the speed of this page rather than homepage;
http://www.pretavoir.co.uk/sunglasses/ray-ban-sunglasses.html
-
Hi Sean,
Your system looks pretty decent... I guess it depends on your traffic load though.
Have you checked out your results on GTMetrix.com?
You scored well in the PageSpeed testing, but not in YSlow. I'd take a look and see if you could leverage some of the suggestions there.
Hope this helps.
Mike
Got a burning SEO question?
Subscribe to Moz Pro to gain full access to Q&A, answer questions, and ask your own.
Browse Questions
Explore more categories
-
Moz Tools
Chat with the community about the Moz tools.
-
SEO Tactics
Discuss the SEO process with fellow marketers
-
Community
Discuss industry events, jobs, and news!
-
Digital Marketing
Chat about tactics outside of SEO
-
Research & Trends
Dive into research and trends in the search industry.
-
Support
Connect on product support and feature requests.
Related Questions
-
Many SEO tools donot allow me to check my site with https?
Hello All, Many seo tools I am using in that few tools do not allow me to test my site with https but it force me to check with http and then give report. So my query is do these tools give accurate report with http same like https? Thanks!
Competitive Research | | wright3350 -
Can deceptive ads help sites to rank? Pogosticking effect involved...
Hello,
Competitive Research | | fablau
I have a question for you. I have noticed that many music sites that appear in the first top results for keywords our users search for, include all the same ads that deceive people by inviting them to "Download" or get the "Full Album" of what people may be search for. Look at the following websites that are often displayed at the first spots for keywords like "Christmas Sheet Music" (just an example): [....] They all have, most of the times, the same ads. Look at the following screenshots taken from the sites above (the ads are at the top of the page): http://www.virtualsheetmusic.com/storage/google/Shot1.jpg
http://www.virtualsheetmusic.com/storage/google/Shot2.jpg Those deceptive ads could engage users more than legitimate websites not showing those ads or not showing ads at all, making Google think that people are more engaged on those website showing those deceptive ads. If it is quality that Google is looking for, shouldn't they do something to avoid having sites ranking well just because of some deceptive ads that take users into action, but without any useful result? I am eager to know your ideas on this issue. Thanks! Fab.0 -
Looking for traffic estimator for specific pages or subdomains on a site I do not control
I'm looking for a free traffic estimator that allows me to look at monthly visits for specific pages or subdomains on a site I do not control. So far I've found trafficestimate.com, but that only lets me look at domains. I'd like to see estimates for subdomains or specific pages. Does this even exist?
Competitive Research | | JGoldberg0 -
Finding new links to site
Is there a way to find sites that linked to my site during a certain time period? Say in the past 2 weeks or from March 1-March 15th ?
Competitive Research | | priceseo0 -
Confused about sites ranking when compared to each other
I'm looking at the keyword "snow guards" for two competitor websites: snogem dot com (SG) snoshield dot com (SS) I've been watching these two for about 4 months. SG stays in the top 5 SERPs consistently, while SS bounces around from 15-20 on Google. I've looked at the competitive link comparison and SS appears to have a much better link profile when it comes to DA. Page specific metrics in OSE show SS with a higher mozrank, moztrust, internal followed links, linking root domains, total linking root domains, and linking C Blocks. It appears to me that SG has many more links, but their links seem to be many site wide exact match keyword links, and paid for advertisements on a major site in the metal roofing industry. Honestly the linking profile of SG surprises me that they rank so much higher than SS after the Panda/Penguin updates. I have two questions: 1. When competing against these sites, should I model SG's link profile? I'm nervous to because of my previous statement thinking they will get hit with Panda/Penguin like updates. 2. What would it take to bounce SS to competing with SG on the first page? Is the number of links the only thing this site is missing, or is there something else to focus on.
Competitive Research | | kadesmith0 -
How can I find the pages on my site with the highest page rank?
I have a lot pages on my site and I'm wondering if there is a tool out there that will find all the pages with the highest pagerank for me? Also, it would be helpful to find that on competitors sites as well. Any ideas?
Competitive Research | | shawn810 -
Sites With Duplicate Content Ranking Way Higher
The site I am writing about is easendtorontohomes.com I don't get it. I see that some of my competitors have more inbound links and have been around longer. Their SEOmoz trust etc is also higher. But I don't get how many could possibly be ranking higher since they're sites are all template based site via Web Tech Design and the content is duplicated from one to the other. The other weird thing is that they do everything wrong - they stuff their keywords tag, some of them use the same keywords/meta data on every single page, some of them don't even have meta tags, some of them don't even have a blog with ANY original content. Apart from listings EastEndTorontoHomes.com has TONS of original, well-researched, and keyword rich content. I just don't get it. Is it possible that google has some sort of relationship with that company? Or am I totally missing something? Some of these sites include: tinasmith.ca, mikeclarke.com, teamkassen.com
Competitive Research | | annasus0 -
Keywords in URL structure for Large Site
We are creating a national real estate website which will contain 2mil+ listings. Our aim is to compete with the National leaders in the field. We are trying to lock down a url structure and it basically comes down to if we should put a major keyword in the url structure or not. The structure would have regional keywords naturally. Its the addition of a more descriptive key word which is in question. Our domain name has no keywords in it. For the sake of this example, we would be targeting "city real estate"... oursite.com/real-estate/state/city/ or oursite.com/state/city/ Here is what the big sites use as an example: http://www.zillow.com/homes/for_sale/New-York-NY/ http://www.trulia.com/NY/New_York/ http://www.trulia.com/real_estate/New_York-New_York/ http://www.realtor.com/realestateandhomes-search/New-York Id love to not have to throw keywords in there and be as clean as possible but i also dont want to shoot myself in the foot. The big boys do add additional keywords in the url, does that mean we should? Id much rather be a leader than a follower but again, I dont want to mess this up from the start and these guys have probably tested this (or have they?). Input would be greatly appreciated.
Competitive Research | | cobbsfriedman0