I have a Concrete5 site which already has a bunch of contents and I want to point the images to my cookieless domain without replacing the urls.
I created an htaccess that will redirect all images from my main site to the cookieless domain
http://www.example.com/images/header.jpg
to
http://static.example.com/images/header.jpg
It's actually working but YSlow doesn't seem to honor this. It's still giving me a low score on that part.
Since you didn't change the image links, browsers will still make a request to the original URL and will send the cookies. That's probably why YSlow is still giving you a poor score for that.
To properly change it you would need to:
Change all links to the new cookie-less domain (static.example.com)
Change cookies to be issued for www.example.com only (per Croises comment above)
Remove the redirects for images
It's a lot of work to achieve, and depending on your site traffic it may not be worth it. Like all YSlow rules (and those from other tools), it's important to understand the recommendations. Not all of them are worth the effort for all websites.
Reference: Cookie-less domains best practices
Related
We use the Clear-Site-Data header for cleaning cookies and other data on logout.
If this header is delivered with the response at https://example.com/clear-cookies, all cookies on the same domain https://example.com and any subdomains (like https://stage.example.com, etc), will be cleared out.
Now this is a problem, if other systems run on subdomains.
I also wonder what happens for domains like "example.co.uk" - would that mean that all the data for all UK company domains is cleared?
How does the browser determine which domains to clear? And can it be influcened?
We want to use it on oursystem.example.org so that it clears only oursystem.example.org and not othersystem.example.org
After looking into the w3c specification it seems like the Clear-Site-Data will clear only the specified domain including it subdomains. So if you clear example.co.uk it will also clear system.example.co.uk. If you clear anothersystem.example.co.uk it will also clear a.anothersystem.example.co.uk and all its subdomains. Here is the example that helped me understand it.
So if you clear oursystem.example.org it will have NO effects on othersystem.example.org but it will have effects on a.oursystem.example.org.
I have around 15+ sites, and we want to drop these sites and merge them into only one site (create pages for each one in the new site).
The 15+ site's domains should be redirected to one site as below:
a.com -> z.com/a
b.com -> z.com/b
c.com -> z.com/c
..
Also, we want to redirect (301) page by page from the old domains to the new the new domain to keep the page's ranking:
a.com/about-us -> z.com/a/about-us
b.com/about/abouus -> z.com/b/about-s
c.com/contactus -> z.com/c/contact-us
Each one of the 15+ sites is running on its own server with a different platform while the new server is IIS.
Currently, I'm thinking of two approaches:
Point the old DNS records to the new server of z.com, and handle all the redirects on the server.
Keep the old site running, and configure redirect rules on each server to redirect each page to the matching page on the new site.
Which approach is better, any other approaches? So far I think the first approach is better since we will control all the redirects in one place - but from the performance wise, is it going to add more headache on the server?
The Scenario
You created a new site and you want to redirect all old sites(15+) to new site (page to page).
All old sites are running in it's own server on different platforms whereas the new one is IIS.
Your options
Point the old DNS records to the new server of z.com, and handle all the redirects on the server.
Advantage
Less cost - you can remove all old sites and the server cost can be saved.
Internal redirection. So, time will be less
Disadvantage
Complex to perform
take care of conflict of similar pages(double check the redirection path)
Keep the old site running, and configure redirect rules on each server to redirect each page to the matching page on the new site
If you can afford the cost of running 15 different sites on 15 servers just to redirect, then only go for this method.
It is just a waste of money and the redirection time will be more.
I think eventually you'll have to shutdown all the other sites as in the long term unlikely does it make sense to keep 15 sites running just to do the redirects.
So as I understand the question is rather on how to better organize the migration to the new system in the short term. So here are my thoughts on this:
how huge is your system
what's your QPS?
how many pages do you have across your sites farm?
do you need to remap URLS for a decent amount of pages?
what's the migration procedure? Will you switch your sites one-by-one or it's technically infeasible and they all need to be swtiched over at once?
If we're talking about a system handling 10 QPS and 1K pages or about a system handling
50K QPS and having 1B pages we need to dynamically remap system load may be a concern and p.2 may look better
rollbacks
note that DNS records can be cached by intermediate servers and if you need to quickly rollback to the previous version if something goes wrong it can be an issue
what kind of systems do you have
Is it actually possible to easily extract URLs from 15 diverse systems and put them to a single point without a risk to lose something valuable?
ease of maintenance
At first glance the first approach looks easier from the maintenance perspective, but I don't know what kind of system you use and how complex the redirection rules need to be.
If they are complex dynamic ones like a.com/product.php?id=1 => z.com/a/iPhone6S moving millions of such urls to a single point could be tricky
SEO
I don't follow the industry closely, but a few years ago both would work ok. I think it's worth consulting those keeping up to date with this industry - it changes very rapidly
Your first approach is definitely the best.
It is easy to maintain
You needn't to keep old infrastructure (though in your second case you'll need to keep only redirecting frontend like apache, nginx or lighthttpd)
There are no performance risks as from one visitor request to old location, redirect answer and than request to new location will come in turn and not simultaneously.
DNS records are not capable of HTTP redirection which is crucial for SEO. To make sure your redirect is 301 HTTP redirect you can use sniffer.
The answer is just make sure your redirect is 301 HTTP redirect so you get your SEO right. Other than that it's just a matter of taste / architecture / money rather than standards.
UPDATE
Read more:
wikipedia
Both Bing and Google recommend using a 301 redirect to change the URL of a page as it is shown in search engine results.
ehow.com
A 301 redirect is a search-engine friendly way to move a domain. The 301-redirected domain does not cause duplicate content in the search engines so that you do not harm your search engine rank. Using a new DNS setting is required to have a new domain name, but it does not redirect browsers or search engines. Both of these methods are used to move to a new domain.
webmasters.stackexchange.com
Duplicate content occurs when the same content is available on two different URLs. To prevent duplicate content on www vs no-www, use 301 redirects to redirect one to the other. To implement redirects, it is the webserver that needs to be configured properly. As long as DNS is pointing to the webserver (either CNAME or A record), then the webserver can be configured properly.
I was stuck with this problem some months back. I wanted to redirect a whole site into a new site's structure. The old site was .php which I know nothing about.
I figured I'd point the old website's DNS at my server and write some MVC code to catch every request, and then use a set of rules using the vb.net Like operator to compare the inbound Url with my ruleset.
It worked a treat. I redirect 300+ pages to my new site with about 10 rules. These include changes of folder structure, a forum (which was mainly junk but had a few good questions), and I implement a "catchall" rule which points to the new home page, in case I missed something.
It worked so well I've packaged it up as a commercial product and it publically available. It is free with link from the destination site (in your case just the single destination site).
https://301redirect.website/
There are a couple of demo videos on the homepage which will explain the setup in a few minutes.
I would like to write a redirect to avoid cookies being sent on graphics & css files. I think what I want is to redirect html and php to www, and others to root, possibly keeping js on the www so scripts can process cookies. This is for Joomla installations that are not cookie aware and I don't want to have to change the template files etc. Related question, can I just redirect the no-cookie files to root if the html is sent to www, or do I need to create a subdomain (which would complicate the no-change policy for the templates)
Thanks.
For reference, here's another SO question along the same lines: .htaccess, YSlow, and “Use cookie-free domains”.
As the accepted answer in that question mentions, creating a redirect from a cookie domain to non-cookie domain would be counterproductive and result in extra round-trips.
I'm not familiar with Joomla, but if as you mentioned the goal is to not mess with the Joomla templates too much, you could do one of:
Register a new domain which is an alias (cname) to your original domain. For example if you already have www.example.com, register examplestatic.com and set it to point to www.example.com. Then adjust your templates to include static files from examplestatic.com. Those requests should be cookie-free.
Use Amazon CloudFront as a CDN. You would use their Custom Origin feature to pull files from your server as the origin. Then adjust your templates to refer to the CloudFront domain instead of yours.
Going down this path may or may not provide much benefit for your situation. You didn't mention it, but I would make sure to start with the higher impact performance rules like minimizing HTTP requests by combining static files, enabling gzip compression, optimizing images, and so on.
my company is running IIS and DNN (I'm not a server guy, so color me ignorant), and I've read previous that you should either redirect your .http://www.mydomain to .http://mydomain or Vice Versa. Can anyone give me reasons to do this? (periods "prepended" to remove href)
From what I understand, it's because search engines see those as two different 'sites' (Even when visiting one or the other, I can be logged into one but not the other).
I also heard it can be a duplicate content problem, which search engines dislike.
Just looking for some professional insight, will help me and others.
Thanks!
This allows your site to be more SEO-friendly. Search engine crawlers will view these as two different URLs. That will cause your site's ranking in search engines to have multiple rankings for the same content.
ScottGu describes the problem and how to go about fixing it in a blog post:
http://weblogs.asp.net/scottgu/archive/2010/04/20/tip-trick-fix-common-seo-problems-using-the-url-rewrite-extension.aspx
Although it's mostly for SEO, there is also a potential usability issue in that a user who logs in on www.domain.com may get cookies that only work on the www subdomain and will be forced to log in again if they ever follow a link to domain.com (without the www prefix).
In addition to the SEO-friendlyness this also prevents some errors that might come up when both, with and without www works.
for example a user could login on www.yourdomain.com and would receive a cookie. later he visit your site via yourdomain.com and the cookie would not apply there.
Background: I've got a web app on sub.domain.com. My primary website is on domain.com. My sub.domain.com pages are stuffed with keywords that I would like to use to get upped in pagerank on domain.com. however, the whole app has been written on sub.domain.com, and it'll be some effort to host it at domain.com/subdirectory, due to how URLs are written, etc.
First question: would you expect that migration (from sub.domain.com to domain.com/subdirectory) to substantially improve the pagerank of domain.com over how it is now? I've done a lot of research, and opinions are split on if google with link the subdomain with the main domain.
Next question: if I do want to do the migration, it'll be difficult to do in the actual codebase (more tedious than difficult). Does anybody have some advice for how I could do this with mod_rewrite? I know there has to be a clever way to do it, but I can't even start to sketch out a solution. Maybe this means it's not a good thing to do, but I was hoping for kind of a quick hack, rather than rewriting all my URLs. Plus, I would like it to be pretty easily reversible, which wouldn't be the case if I change my URLs (dev is ongoing, so it's not as simple as just rolling out a previous version).
Pagerank isn't a property of domains, it's a property of individual documents. So it'd be more accurate to say that migration from sub.domain.com to domain.com/subdirectory will improve the pagerank of domain.com/subdirectory. If you're concerned solely about the ranking of the domain.com home page, the impact on that will mostly depend on your internal linking. For example, if all pages on sub.domain.com currently have a "home" navigation link that leads to the home page of sub.domain.com, and when you do your move they'll now lead to the home page of domain.com, then this will contribute to the domain.com home page's ranking. If this "home" navigation link went to domain.com/subdirectory, on the other hand, then that's what they'll be contributing pagerank to.
mod_rewrite doesn't change the outbound links in your HTML, it changes how inbound links are interpreted. So it would let you put this in the virtual host file or .htaccess for sub.domain.com:
RewriteEngine on
RedirectRule (.*) http://domain.com/subdirectory/$1 [R=301]
to mass redirect any requests coming in on sub.domain.com where they need to go. It won't help you produce correct new-form URLs in your codebase. (You could, in theory, leave all your links how they are and rely on the 301 redirect to keep you from having to change them, but this is really sloppy and wasteful, generating two HTTP requests instead of one for no good reason).