This blog post describes the improvements that Amazon Web Services (AWS) could do to deliver a better service to customers everyday, solve more problems to them and in the process, generate more revenue for the company.
First Point: Build a excellent feedback system
Amazon has proved that its services are one of the the best in terms of ROI based on the pay-only-what-you-use business model. But What happens if any service goes down in production in any window of time? You know that it actually happened in 2011, when a large quantity of servers of the global Amazon infrastructure had a massive outage. For the customer, it doesn’t matter if it was a electric failure, a massive DDoS attack, or another problem. For the customer what really matters is that the service never can go down.
Amazon, in the last year fixed a lot of problems related to this outage, and everyday it has built more confidence with customers. But what about the problems for certain service? What about if certain AMI type has a minor or major problem? Where the customer should go to report these kind of problems?
So, my thought about this, if that Amazon should build a amazing, reliable and useful feedback system for the reporting and analysis of these problems, or better to create a partnership with the startup Uservoice for the integration of AWS with this system, and generate revenue for both parts.
Second Point: More engagement with the Open Source community
There are many good Open Source events where Amazon doesn’t have a strong presence, and we are loosing the opportunity for key part of the process: networking. LinuxCon, Ubuntu Developer Summit, and the Ubuntu Cloud Summit, FOSDEM,Data Science Summit, OSCON, Surge Conference, etc. If the Amazon’s employees can have a better participation in these events, meet more people, that after could be potential customers, it would be simply awesome.
And another though in this point is to involve more kernel developers on the Linux kernel development, because, the main platform for EC2 is Linux, so Amazon should play a key role in this project, to give the chance to advanced customers to choose the more updated kernel version for them. It was one of the problems that I identified where I was managing some EC2 instances with very old kernels (based on CentOS 6.2). There are many advanced customers that they want to have the choice to install a updated version of the kernel, for example the 3.2 version that have many improvements in the memory management and I/O scheduling, a critical resource for EC2 instances.
Third Point: Build a partnership with OmniTI and create an OmniOS-based AMI for EC2
OmniTI is one of the most respectful firms of the world for Scalability and Performance problems, and if Amazon has this incredible group of experts like partners, it will explote the revenue for the company, because with the expertise of Theo, Leon, Robert and many more fellows, Amazon can build a strong confidence with more customers saying to them:
“We can host you applications in our servers and services, and if you have a scalability trouble, we can work with Theo and his team on it”
OmniTI has built an amazing operating system, like a enterprise version of IllumiOS, the fork of OpenSolaris created by many principal developers of this amazing operating system, after that Oracle closed its sources. They are Solaris’s experts and for that reason, they built a very seemed environment to deliver a better operating system to give to their customers, and the result is OmniOS. If we can build an OmniOS-based AMI for OmniTI customers, they can promote to AWS wherever they go, and we can deliver the best service for OmniTI’s customers. What do you think about this?
Fourth Point: Use the MapR Hadoop M3 distribution for the ElasticMapReduce service
Ted Dunning and his team at MapR Technologies have created the best Hadoop distribution, at least for me, divided in two versions: a free version called M3, and a commercial version with support 24x7x365 called M5 with great features not found in other version like:
- Real-Time Data flows
- Built-int compression
- Multiple clusters support
- MapR Volumens
- MapR HeatMap
and many more.
Look for the MapR site and read the DataSheets for every version, and valuate the amazing features that brings these Hadoop versions. It would be nice to talk with the Elastic MapReduce Product Manager to discuss these points. For example, What about to form a partnership with MapR Technologies with a very seemed schema that they did with Greenplum, one of the leaders today in the Big Data field? What about to create an alternative ElasticMapReduce service based on MapR M3 or MapR M5?
This is a personal opinion what I would do at AWS to create one of the best Cloud Computing platforms for every customer. If you have a comment, idea to discuss, write a comment, or Tweet me to @marcosluis2186, or simply send me a email to firstname.lastname@example.org. Best wishes and thanks a lot for your time.