Before I talk about AntStack, I must talk about how the three founders came together in this journey. I and Jeevan worked together before in a fashion E-Commerce startup, Styletag. I was the CTO and Jeevan was heading DevOps. Together, we shared the journey of evolution of cloud technology. There were a bunch of challenges we faced along, and I would be elaborating them further. These challenges played a pivotal role in brewing the idea of AntStack.
Journey of Styletag
Being an E-Commerce startup, it was tricky for us to predict the traffic. Some marketing campaigns from which we anticipated fair performance would fail, while others would perform exceptionally well without much effort. This made us wonder what were we lacking and how could we foresee the traffic and be prepared for it. Although we had an ELB setup, it needed warming up and our traffic used to come in bursts. This was one of the major challenges we faced with regard to capacity planning.
Monolithic Rails App
Our E-Commerce platform was built on top of Rails gem called Spree Commerce. Like a typical startup, we were shipping new features rapidly and over a period of time, we accumulated huge technical debt. It became harder and harder for us to scale since we reached a point where “throwing servers at it” didn’t help. We had to use APMs like New Relic to optimise our memory issues as much as we could, but that didn’t keep us afloat for long. At one point, we ended up paying more to New Relic than to AWS!
Scaling Database and Over Provisioning
Since the database was scaled vertically and involved downtime, we had to plan it properly. Because of this, we had to keep our db capacity with burst traffic in mind, not average traffic. This resulted in over-provisioning; basically a lot of unused capacity.
Nobody talks about the constant cognitive load created when you are responsible for the uptime. Every “ting” on my phone, secretly made me wish it wasn’t “pager duty”, since I would be constantly thinking of it. Followed by this would be a trail of questions like, is the service down? May be recent prod deployment caused a memory issue? Did I follow the deployment steps properly?…and the list got bigger, not better.
However, Styletag had to be shutdown after an exciting journey of 7 years due to cut-throat competition in E-Commerce space.
Journey of WheelsBox
Soon after my stint at Styletag ended, I started working on WheelsBox along with Latif Hussain, who hails from the Automobile industry, owning car dealership (Blue Hyundai). WheelsBox is a platform which aims at making new car buying experience seamless for end users. I wanted to build WheelsBox with fresh perspective at technology as it didn’t make sense to go through the same journey as Styletag. After all, we should learn from our past they say! I wanted to keep the tech of WheelsBox simple and pull it off with a 4-people team including myself since the budget was a constraint.
Serverless was the go to choice since I didn’t want to get into managing servers ever again! Generous free tier on AWS was also a great thing to get started, as I didn’t have to think about bills while developing the product. We used Serverless Framework to bring up our APIs powered by Lambda Functions (NodeJS) and DynamoDB. It wasn’t easy for me to change my mindset in the initial days and to stay away from my biases to embrace Serverless. There were many days I told myself, *“let me just do it in Rails,”* but, again I used ask myself ”Do you really want to go back to those days?” and kept pushing until I got it right.
Best Practices Enforced from Day-1
Since we were forced to fit our code into Lambda Functions, it was implied that we divide our application into smaller pieces. It was like micro-services minus all the management headache. If there was some bad code (there would always be some), it would only impact that Lambda and not the entire system. It was easy to debug, and in worst cases we would just rewrite that Lambda Function.
Focus on Product, not Infrastructure
Since we went 100% Serverless, there was nothing in our stack which needed monitoring. This was a huge relief since we could use that time to think about product enhancements and not worry about keeping the servers running. The time required for scalability, capacity planning, reserved instances planning, configuring APM, was available to be effectively utilised in enhancing the product.
We managed to develop WheelsBox in about 10 months. It is being tested on staging at this point, soon to be launched to the public.
Journey of AntStack
I was very proud of how we developed WheelsBox and started talking about it with friends, including Jeevan. Jeevan was excited about Serverless since we had shared the struggles during Styletag days. He started talking to people about Serverless & we started seeing a lot of interest from many people to explore it. At the same time, I was explaining Serverless to Latif and being a non-developer, asked me this fundamental question: Why aren’t there many organisations which are making a move to this technology? They can greatly benefit from cost savings & also costs will grow in proportion to the business growth.
It is that question which made us come together and start AntStack, to help other organisations with our learnings and make them leapfrog into Serverless. AntStack’s goal is to free organisations from the grip of technology and instead make technology work for them. In Jeevan’s words “make technology invisible”.
Me & Jeevan thought of getting into Serverless consulting since we saw lot of people asking us to help them out. But, Latif brought in the big picture of what if this can be bigger than that & took care of end to end need of product development in Serverless paradigm. Me and Jeevan agreed with Latif that this can serve bigger purpose than just solving a small piece of product development. With that in mind we started AntStack with following offerings:
- Serverless Architecture & Consulting
- Serverless DevelopmentQA
- Frontend & App Development
- UX & Product Design
- Voice User Interface (VUI)
We wanted to make sure an organisation could leave everything to us & not look at 3 or 4 different set of companies to deliver parts of the product. If you need help in any of the above or all of the above, do get in touch with us & we will be more than happy to help you in this journey.
AntStack - Making Technology Invisible