Tagged: , , ,

Viewing 8 posts - 1 through 8 (of 8 total)
  • Author
    Posts
  • #24554
    Profile photo of Robin
    Robin
    Keymaster

    As we near the final stages of the Symzio engine, we started exploring the end service from a global perspective. Our goal, from the start, was to compete at a level that is on par or superior to the larger agencies. In order to accomplish this for customers, we needed a service that would provide quick results for customers in South Africa as well as Japan, and this cannot be accomplished using the standard single server solution.

    Well, it actually can. Use a CDN.

    We ran a test using Cloudflare’s Free CDN on one of our Symbiostock PLUS sites. Here were the results:

    With Cloudflare disabled:

    With Cloudflare enabled:

    This indicates, on average, a 31% increase in load speed for customers.

    The way a CDN works is, they have servers all across the globe in different places. When a person accesses your site from a part of the world, it delivers content from a server closer to them. So even if your server is in Brazil, a user in Russia will actually be getting your content from a CDN server in Poland, for example.

    Aside from the speed increase, it will reduce the load on your server as cache’d stuff will be loaded from the CDN rather than your original server, as well as provide nearly absolute protection from DDOS attacks as the CDN acts as a shield.

    The CDN also has intelligent algorithms that ensure that dynamic content, such as your cart page, continue to provide non-cache’d content so your dynamic content remains intact.

    In my opinion, every single Symbiostock user should switch to a CDN, and I recommend the free Cloudflare CDN service.

    One additional note – Symzio will be sourcing all its media directly from your sites; it will not save copies of your images (in most cases) on its own server. As a result, we are going to be ordering results on each individual search page according to how fast the image loads for a customer. This means, you are likely to get more exposure and more sales the more globalized your site is. Accordingly, it is a near certainty that Symzio contributors that use a CDN will have a direct advantage over users that don’t.

    How difficult is it to use a CDN?

    All you do is sign up with them and switch the nameservers through your registrar to them. That’s it.

    We are switching every site we run over to a CDN, and see no reason why every Symbiostock user shouldn’t as well.

    #24571
    Profile photo of Andre
    Andre
    Participant

    A CDN would make a caching plugin redundant, wouldn’t it?

    #24572
    Profile photo of Robin
    Robin
    Keymaster

    It depends on the type of caching that is done, but for the most part, yes, it will remove a lot of load from your server so a lot of the local caching becomes irrelevant.

    #25369
    Profile photo of Robin
    Robin
    Keymaster

    Just a note for everyone – after doing some testing, we’ve come to the conclusion that it can sometimes become problematic when trying to access new data – just be aware when using one.

    #26081
    Profile photo of Milan Stojanovic
    Milan Stojanovic
    Participant

    Did anyone compare Cloudfare to Photon (from Jetpack plugin). I’m asking because on many sites it says Photon is really helpful with hosting many images.

    ON their site is says: “Photon currently only acts on images in posts and pages, as well as featured images/post thumbnails via the image_downsize filter.”. Since our symbiostock images are not stored as posts or pages, it means Photon will bypass them?

    #26083
    Profile photo of Robin
    Robin
    Keymaster

    All product images are stored as “featured images/post thumbnails” so yes, it should work.

    #26084

    Done! Pretty good improvement. 🙂

    #26085

    Please delete my previous post. The tests are from different locations which does not justify the result. Sorry for that.

    The website I used for loading speed test is http://tools.pingdom.com. I set the test location to Sweden for all tests and got quite different results ranging from 5s to 0.8s for the same website. If test results themselves are not consistant, I can’t figure out the difference between before and after. Anything missing here?

     

Viewing 8 posts - 1 through 8 (of 8 total)

You must be logged in to reply to this topic.