Feedback for Amazon Web Services
Ingrained in the DNA of the Amazon Technologist is a single-minded focus on the needs of our customers. The Amazon development process is even called "Working from the customer backwards". Essential in this process is a good understanding of what the customers need in terms of new services, new features for existing services, or different approaches to things that we are already doing. We collect this feedback continuously from various sources: the AWS forums, the AWS Premium Support Team, Amazonians on the road talking to customers, solution architects helping to define customer architectures, ISV partners building on our services, system integration partners who relay customer needs, advisory boards, and of course the Amazon ecommerce engineers building on the AWS platform.
Once a year however we take a moment to make sure that everyone who wants to give their input into the direction of the Amazon Web Services has the opportunity to do so. We have developed a Survey that helps us define what is really important to our current and future customers. If you have feedback that you would like to give to the Amazon Web Services team, this Survey would be an excellent place to do so.

That’s true Werner. Being customer centric rather than technologist is our biggest challenge as solutions architects. It’s amazing how working “backwards” simplifies, and speeds up, the development cycle.
Cheers
Luiz
@luizb
Thanks Werner
Now I have an official place to voice why I haven't signed up for CloudFront yet (even though I really, really want to).
It's because of the current lack of:
1. automatic gzip compression (drastically reduces bandwidth - faster download)
2. last-modified http header (allows web browsers to cache images)
Thanks!
Thanks Tim, forwarded your comment also to the Cloudfront team directly.
Following up Tim’s post, you can set HTTP headers with Cloudberry on S3
http://blog.cloudberrylab.com/2009/06/how-to-set-custom-http-headers-for.html
Thanks,
Luiz
@luizb
Supporting customers (and their customers) by providing the basics that traditional ISPs have provided for a decade plus is EXTREMELY important.
- Being able to use multiple IPs per instance is a HUGE deal especially if you need more than 1 ssl cert on an instance (with a different domain name).
- Sticky sessions with load balancing service - HUGE also and i understand you are working on it. Since the very first time we have used a load balancer (over a decade ago) we always employed sticky sessions.
- Reverse DNS so our websites and clients who are trying to offload infrastructure can have SMTP servers that have some chance of getting reputation and/or decent deliverability.
- EBS volumes that are read AND write for multiple mounts. Not just read only and single write (as in a single master db scenario with multiple readers). This does us no good for multiple web front ends all sharing web data.
- Online imaging / bundling with multiple drive support so we can snap shot an instance in a point of time. We should not have to take a server offline or do a bundle and several snapshots and scripting to somehow label them for a disaster recovery that may or may not happen.
- Windows love. If you don't give windows hosters love soon MS will take them from you with Azure. You know it won't take them much longer and we don't even have windows 2008 support or reserved instances. Some guidance for creating instances preloaded with sql 2008 web is huge as well!
Our clients are paying substantially more than they did with their previous hosting arrangements when the intent was to provide maximum scalability with lower cost. Due to the issues above this has not materialized. We are being pressured to move to other cloud providers who have more windows centric offerings and are frankly cheaper because of it. Help us out! We want to stay with AWS!