Flux7 CTO Ali Hussain Analyzes AWS’ News from re:Invent 2016
This month’s re:Invent in Las Vegas drew over 32,000 attendees and the show did not disappoint as AWS delivered on its precedent to unveil a number of new features and products at the show. With numerous announcements, AWS news was peppered throughout two days of lengthy keynote sessions, we’ve asked Ali Hussain, Flux7 co-founder and CTO, to weigh in on what caught his attention and where he thinks the most impact will be seen to enterprise organizations like those that Flux7 serves.
Q: There were several announcements across two days of keynotes. What stood out to you most and why?
A: One of the first things that caught my attending was AWS Greengrass. It seems very valuable as unreliable internet is one of the biggest issues organizations face with their IoT initiatives. A great example of this is the work we did for Fugro where its IoT devices collect critical data on ships that can be located anywhere in the world. To transmit the collected data, Fugro devices often had to communicate through unreliable satellite internet with expensive data transfer costs.
In scenarios such as this with a hostile internet environment, you need a software strategy that allows the devices to continue to work when internet is not available. This presents a major software development problem. And while different paths can be taken in different scenarios, this implies different copies of code which not only has a distinct development cost but an even more expensive maintenance cost.
By letting you use the same software development model and platform, Amazon provides a clean process for separating the business logic from the plumbing. Already you have the business logic deployed with AWS Lambda when you have access to the internet but Greengrass allows you to seamlessly use the same business logic when you don’t have an internet connection.
Also, as someone with a hardware background, I found the FPGA instances to be very interesting and think they can be extremely impactful. For specific problems using specialized hardware, FPGA can make orders of magnitude of difference. For some time now, Amazon has recognized the importance of specialized hardware. For example, they have long had GPU instances. However, FPGAs allowing specialized software can really take things to the next level.
To understand the great difference here, compare bitcoin mining against using general purpose hardware with specialized ASICs. There’s a 10000x difference between CPUs and ASICs and 1000x of GPU vs ASIC in hashes per dollar. Indeed, the FPGA instances are a very interesting offering for processor designers, several of whom Amazon has close relationships with.
Q: Anything else stand out to you?
A: For AWS SnowMobile, we expect that to be used by our oil and gas customers who have mapping data that needs to be analyzed in the exabytes. SnowMobile will allow them to bring those applications to AWS and analyze them at cloud scale. One of our bioinformatics clients, Personal Peptides (aka. ImmunoDiet) wanted to process genomics data in AWS. One of their challenges was the time it took to import all the data from nrDB. We are expecting that over time Amazon’s focus on data import will make it easier and easier for customers who need to import large amounts of data.
Q: There was visible excitement in the crowd when AWS announced its PostgreSQL for Aurora. Do you have initial thoughts on the implications of this announcement?
A: AWS Aurora takes database management to the next level. It is very scalable, offers high availability, and great fault tolerance. Before this announcement, only MySQL users could take advantage of the extra features offered by Aurora. Now Amazon changed that.
This is a highly-anticipated feature. Postgres has a very loyal following with it being the favored choice of database among many platforms like Heroku and ISVs like Atlassian. Favored in financial environments for its reputation of reliability, several Flux7 customers can immediately start using this new feature.
We are working with a local Austin-based company who helps organizations securely and effectively monitor threats, streamline notifications, and recover from incidents who is currently using PgSQL as their primary DB in AWS. As a company that has a large development team and no dedicated IT teams, they are always looking for ways to reduce their ongoing maintenance. The Aurora database has caught their attention as it has a reputation of being really fast and really easy to manage. Due to this, they had a task on their 2017 technology roadmap where their developers were going to change their code to use MySQL instead of PgSQL. With this announcement, they were able to cross that major task off their to do list in 2017. We will soon be moving them over to Aurora now without any code changes necessary.
Q: What did you think of the deep learning announcements around Lex and Rekognition?
A: Amazon LEX and Rekognition represent a real peek into the future. The primitives we work with are rapidly changing. By adding best in class image and voice recognition, businesses won’t have to re:Invent the wheel, pun intended. This is particularly true for some exceptionally difficult problems where organizations would rather focus their effort on how to use these tools to provide value to their customers.
Q: How do you think CodeBuild will benefit Flux7 customers?
A: We are particularly excited about CodeBuild as we were in the private beta and have been playing with it for some time. It has clear potential. It will be an extremely useful tool as it allows us to have a completely serverless build pipeline. Developers push to AWS CodeCommit, CodePipeline triggers a build on CodeBuild and uses CodeDeploy to deploy the build and promote the code up the pipeline. Just within the last two weeks, we have already designed an architecture (being implemented now) for a loyalty program company which uses code build not just for delivery of code but also for CloudFormation and Ansible playbooks.
Q: When Pat Gelsinger of VMware joined Andy Jassy on stage, they shared several anecdotes of customers quite excited at the news of being able to run the same VMware software on AWS that they run in their data centers. Are you seeing the same reaction in the market and do you expect enterprises to quickly take advantage of this?
A: When the VMWare announcement came out a few weeks ago, it caused a lot of buzz as it represents a few things. First, it makes it very easy for VMWare users, (which is almost everyone with a data center), to simplify migration to AWS. This is going to accelerate the cloud migration effort for a large number of our customers. Similar to how the RackSpace announcement last year made it clear that RackSpace sees more value in being aligned with AWS than to compete with them, this news validates that the Cloud is the future and AWS is leading the way.
At Flux7, we see it as a very welcomed change. As a company focused on causing enterprise transformations, lift-and-shift migrations are always something we need to do but not the task we prefer to spend a lot of cycles on as it does not contribute deeply to the transformation. By lowering the barrier to entry for lift-and-shift into AWS, this partnership has effectively enabled us to focus more on our speciality and spend less time on the necessary pre-requisites.
Q: Any parting thoughts?
A: Andy Jassy said in conclusion that the next ten years of innovation will be fueled by those organizations who are laser focused on the best way to transform their existing business and build new business opportunities with the cloud. At Flux7, our business model is wholly aligned with that sentiment, helping businesses assess, plan for, design and implement IT transformation that leads business success. Aligning with this sentiment, we have a philosophy that we teach our clients how to fish, and it is truly exciting to be the Sherpa who gets to witness the next decade of innovation. As testament to that, this is the third year in a row, that Flux7 customers have been featured in AWS keynotes and we look forward to many more.