I have recently deployed my website to a server located in the US. I’m sitting in Europe, so I was not thrilled by the performance. It took a long time to load the site, which meant that people would not say on my site.
I therefore was wondering what I could do to improve performance. Since I’m using Amazon Webservices for other tasks in my business I came across Amazon Cloudfront .
Cloudfront is a services, which can be used to serve static content to users from a site closer to them self. Cloudfront currently serve data from 4 datacenter in US, Europa, Hong Kong and Japan. The distance from the user to the server is therefore reduced, making response times lower. This is important when the sites contain a lot of different images, which have to be downloaded to show the site correct.
My website is a CMS system running on an Apache and PHP server. The method can also be used to improve performance for SAP Netweaver portals. The portal also contains a lot of static content like images, style sheets (CSS) and javascript files. It might therefore be a way to improve performance or impression for the users, if the content was placed on a server closer to the users. I’m not the familiar with the definition of the Webdynpro environment to know where to change the location of images or java scripts, but believe it could be possible.
It seems like a difficult process to get started with using Cloudfront, because the need of finding the correct API’s. Luckily I found a nice guide on setting it all up at http://www.hongkiat.com/blog/amazon-cloudfront-how-to-setup-cloudfront-to-work-with-s3/. This blog introduces two tools.
- A plug-in to Firefox, which make it fairly easy to upload and share the content. This plugin can also upload content to Amazon S3, which is a storage service.
- The other is a tool, where you can test load time of your server. This tool can be used to validate it the loading process improved or and getting an overview of which images is loaded by which scripts. The tool is located at http://tools.pingdom.com/.
Before I started improving the service my load time was 8.5, and the load profile looked like the following.
After I had performed the improvement the load time was just 3.3 second.
I now have a much faster website, which did not require a lot of investments.
1 Comment
