Top 5 Sugar Lessons Learned from AWS re:Invent 2016
Amazon Web Service’s re:Invent conference is one of the leading events for showcasing innovations in cloud technology. This is a particularly exciting time for us, as we have just completed moving all our computing infrastructure and data to Amazon’s cloud servers. This year, as Director of Technology, I — along with more than 30,000 attendees — went to AWS re:Invent to check out some of the hottest products displayed at the show, while engaging in a plethora of valuable education and networking opportunities.
In this post, I’ll highlight some important takeaways from the event that can be applied to Sugar hosted companies.
#1 - Small changes can make a big difference.
The biggest epiphany I had was during a session on EC2 performance. EC2 is the elastic compute portion of Amazon, centering around virtual machines that live on top of physical machines. In this session, the presenter talked about things that we would have never thought of in a billion years to increase the performance of a machine, such as changing how the machine gets its reference of time. It is best practice to change the timekeeping settings from the old default Xen clock source to instead use the TSC clock source available on new instances using the Intel SandyBridge chipset. Timekeeping calls see a performance improvement upwards of 40%.
You would not think that this would be a big deal, but it can lead to higher performance. We were given a host of small changes that can be made to create a huge impact on performance.
#2 - Dynamic scaling is key.
Amazon is all about trying to help customers save money by spending less to get more. Which is a totally strange concept for us as consumers to understand because people usually want more money for more services. But here Amazon is saying no, what you should be doing is starting very small and then using various parts of our service to scale up when the need arises. Then, if that need goes away, your service should automatically scale back down to a smaller size, so that you don’t waste money on resources that are idle. This is counterintuitive from an IT perspective. Traditionally, you plan for a maximum and deploy resources to meet that maximum. Amazon is offering a concept best referred to as “computing at scale”. In Sugar terms, this could allow you to add web servers as more user connect to the system or add job processing servers as scheduled jobs consume resources. When the user or job load is less, we can automatically remove these machines
#3 - One size doesn’t fit all.
There is no single out of the box configuration that will fit every workload. Some SugarCRM customers have more read intensive operations, while other have more write intensive operations. Some customers may require a secure connection back to the home office to pull in back office data. Here is where Amazon shines. There are many configuration options that can be employed to optimize a particular deployment for a specific need. Additionally, Amazon offers very robust networking tools so that you can make your servers in Amazon an extension of your existing networking infrastructure.
#4 - Linux is king. If you are hosting your Sugar instance in Amazon, you should be hosting it in Linux. All the best performance improvements that you can make are easiest to make in Linux.
#5 - Security rules.
In every session attended at AWS, security was one of the first and foremost things talked about. Security is vital; you should be protecting your systems not only from people on the outside but even people that might work within your organization or people that work at Amazon. Amazon has many methods for encrypting data ‘at rest’ that we will be deploying for ourselves and our customers in 2017. AWS Key Management Services (KMS) is supported throughout Amazon services to encrypt various types of data. If you want to meet strict standards such as HIPAA, you will need to implement data at rest encryption.
Attending AWS re:Invent was like an experiment in controlled chaos. There were tons of people, from C-Level to marketers to developers, all reveling in the endless opportunities for learning and networking with their peers. We at UpCurve Cloud learned a lot and are excited about how we can apply these lessons to our customers, leveraging our knowledge of server operating systems and Sugar to optimize performance for the end user.
Contact us to learn more about how UpCurve Cloud can utilize AWS to optimize the performance of your CRM platform.