Do services like Cloudflare and Incapsula actually improve the performance of websites hosted on Windows Azure?

Question

I'm running an image-heavy website hosted on Windows Azure. Back-end performance is great but response times for image thumbnails, which make the bulk of page sizes, are quite volatile. I'm using the Azure CDN for serving all images but their response times vary by orders of magnitude and I haven't found any pattern in the fast (~150 milliseconds) vs slow (3-4 seconds) requests yet. This also doesn't seem to be a local phenomenon since I've tested the load times from different locations/continents. My conclusion so far is that the Azure CDN is simply not that good after all and I started looking for other ways to improve the load times of static assets.

Now that the context is clear, here is my actual question: does anyone have experience with services like Cloudflare and Incapsula for improving the performance of websites hosted on cloud infrastructure like Windows Azure? These services promise reduced server load among other things, but I'm more interested if they are actually effective in reducing response times for static files, as well as any negative impact on dynamic page content. I'd greatly appreciate any answers based on practical experience and/or advice for alternative solutions.

UPDATE: Here are the response headers for one of the images on the CDN:

HTTP/1.1 200 OK
Cache-Control: public, max-age:31536000
Content-Length: 4245
Content-Type: image/jpeg
Last-Modified: Sat, 21 Jan 2012 12:14:33 GMT
ETag: 0x8CEA64D5EC55FB6
Server: Windows-Azure-Blob/1.0 Microsoft-HTTPAPI/2.0
x-ms-request-id: d7a1ef38-6c99-4b38-a9f5-987419df5d24
x-ms-version: 2009-09-19
x-ms-lease-status: unlocked
x-ms-blob-type: BlockBlob
Date: Sun, 05 Feb 2012 12:56:12 GMT
Connection: keep-alive

Answer 1

Incapsula has two caching modes: 1) Basic - this mode caches static content according to directives in the http headers (in the same way a browser would or a commercial caching proxy would behave). This typically provides 30%-50% improvement 2) Advanced - this mode also caches static content that was not specified in the http headers and dynamic content by using advanced learning capabilities to determine what content is cache-able and when to expire the cache. These methods are optimized for striking the right balance between utmost caching and serving fresh/up to date content. This mode typically adds an additional 20%-30% improvement.