I've recently launched a new site which makes extensive use of the builtin Image Cropper data type however over the weekend the site experienced massive issues when under high load and these were undoubtedly down to ImageProcessor serving the crops.

The master template of the site contained approximately 100 calls to GetCropUrl which under load resulted in CPU utilisation of 100% and at one point a memory usage of 8.5GB for the Worker. When the elements of the page containing the calls for crops were removed from the template, all issues on the site stopped and CPU & Memory usage returned to normal levels.

Umbraco version is 7.4.1 running on a server with 4 CPU cores and 24GB of memory. ImageProcessor version is 2.3.3 and ImageProcessor.Web is 4.5.3 which I believe are the latest versions.

I believe that these issues could be prevented by removing the need for ImageProcessor and just having the Image Cropper create the crop image at document publish. This was the behaviour of the original Image Cropper so I'm unsure as to why it has been changed, or if it's possible to update the cropper so this functionality can be reenabled?

Thanks

Paul