When a CEO learns his company will appear on the television show “Shark Tank,” the natural reaction is excitement for potential hockey stick growth. A CTO’s reaction? “Oh no.”
Adam Berlinsky-Schnine, CTO of Apairi, wrote about that exact experience in a blog post for Hacker Noon. In it, he sums up a common scalability fear.
“The main reason scaling is so stressful is that it is not linear: your system could be functioning perfectly one minute, then a small increase in usage tips it over the edge and you have an outage,” he said.
Engineers across New York City are looking for how to best scale their own tech stacks without disruption to customers or drainage of funds. When companies are ready to scale, part of that process means outgrowing old systems. Director of Engineering Noah Appel at INTURN, a company that turns excess inventory into capital, said they did a whole overhaul to their core SaaS product in order to implement scalability. Notably, they transitioned their front end from clunky proprietary systems to community-supported, open-source libraries and frameworks. And CTO Bernard Kowalski said home insurance marketplace Young Alfred transitioned to the cloud for infinite scalability.
After deciding on tech tools, team processes need to be implemented. How are engineers prepared for larger data responsibilities, and what are the playbooks if a mistake or outage happens? Plan out time for tech training and adjustments. When it’s time to scale, the team will be ready.
CTO Bernard Kowalski said scalability is a marriage of tech tools and team processes. It’s difficult to add more servers, CPUs and memory without implementing best practices to handle the workload. At home insurance marketplace Young Alfred, scaling in the cloud means constant monitoring, logging and alerting.
Describe what scalability means to you. Why is scalability important for the technology you're building?
There are two primary factors that drive scalability. First is software, design decisions and IT infrastructure. The second is the scalability of teams and processes. It is difficult to build scalable systems without experienced engineers tuning both parts of the engine.
Scaling can also be vertical or horizontal. We usually think about horizontal scaling, where we add more nodes to the system to handle additional work, instead of vertical scaling, where we just throw more memory, more or faster CPUs and more storage onto individual machines.
Vertical scaling is easy if you have more dollars initially, but it has hard limits. To design systems that are expected to handle orders of magnitude more work, planning for horizontal scalability is crucial. The ability to quickly increase processing capacity and manage costs is especially important for startups on an exponential growth trajectory.
How do you build this tech with scalability in mind?
Cloud is the way to go when it comes to scalability. For an average startup, it provides almost infinite scalability for properly designed systems.
Cloud is the way to go when it comes to scalability.”
What tools or technologies does your team use to support scalability, and why?
Microservices, together with architectural patterns like CQRS or event sourcing, built on the modern cloud infrastructure help with scale because you break large complex problems into more manageable pieces. SaaS services like ECS, Kubernetes, elastic storage, CDNs, load balancers, data lakes and more enable software engineers and architects to build systems that can naturally scale. Docker is also a game-changer for many software engineers.
Having all these technologies at hand is great, but combining too many can increase complexity. We find ourselves constantly balancing the complexity trade-off.
The creation process of a system that is scalable is only half of the success; operating at scale is the other half. We must be able to diagnose and fix application issues that arise from the underlying infrastructure while meeting SLA requirements.
This is where we need to follow best practices of building for the cloud with proper logging, monitoring and alerting. On the testing side, lower environments that are as close as possible to production are very important. We use canary or blue/green release strategies, where you can safely run new versions of your applications while minimizing the risk of a negative customer experience.
Finally, approaches like domain-driven design help reduce architectural complexity. Having a proper CI/CD pipeline significantly reduces pain points with the development and deployment of multiple interconnected services. It is a best practice to automate as much as you can in all aspects of the development process.
At real estate tech company Common, VP of Operations Eric Rodriguez said automation helps them scale while also reducing costs. From there, the engineering team can focus on critical issues.
Describe what scalability means to you. Why is scalability important for the technology you're building?
Scalability is about being able to do a lot more with a lot less. Scaling is a critical piece of the operations of any company. You need to think constantly about actions that can be automated to allow you to reduce costs but also to make sure that you have your employees focused on the most critical issues for your business.
As Common looks to build a resident brand and the next generation of property management, we need to make sure that we’re scaling effectively to reduce costs for the owners and developers of the properties that we manage. This allows us to provide a delightful and mostly self-serve experience to our members at any time.
Scalability is about being able to do a lot more with a lot less.”
How do you build this tech with scalability in mind?
As an operator, any time I work with a development team to think through building tech, I like to first outline each step of the process in a flow. Once the process is diagrammed, I push myself and others to identify what parts of the process we think can be automated. A frequent question I ask is “Why can it not be automated?”
That question is the foundation for scalability and allows us to identify places where automation might be detrimental to the experience of our customers as well as drive alignment on the path forward. Another thing I like to think about is how to make a tool be as self-serving as possible so we don’t need to keep bothering the engineering team to make tweaks as we learn more about user behavior or as the scope expands. This isn’t always possible, but it’s definitely something important to keep in mind.
What tools or technologies does your team use to support scalability, and why?
Our operations team relies on Salesforce pretty heavily, given that it’s a platform that can be built to meet your team’s specific needs and integrates with a lot of different third-party tools. Through Salesforce, we’ve been able to automate different points of communication in our customer journey and have automated the scheduling and carrying out of tasks across our funnel. We’re now even starting to explore robotic process automation to help us automate any lengthy manual task the team has to carry out.
Director of Engineering Noah Appel said scalability at INTURN, a company that turns excess inventory into capital, requires onboarding training for their clients’ engineers as they scale. To enhance the customer experience, Appel’s team built new languages, frameworks and philosophies into their tech stack.
Describe what scalability means to you. Why is scalability important for the technology you’re building?
Scalability in the B2B SaaS world requires the ability to onboard, support and satisfy larger companies that need to transact on larger and larger datasets. It also means the ability to onboard new engineers as our own engineering team grows to meet the needs of larger enterprise customers. As we begin to onboard enterprise fashion and CPG brands, this ability to ingest, manipulate and read massive amounts of data while providing an exceptional user experience places the challenge of scalability front and center.
Scalability in the B2B SaaS world requires the ability to onboard, support and satisfy larger companies.”
How do you build this tech with scalability in mind?
In our overhaul of our core SaaS product, starting with scalability in mind, we transitioned our back end from a monolithic to a microservices application and transitioned our front end from clunky proprietary systems to community-supported, open-source libraries and frameworks. We are also rebuilding our DevOps systems to increase self-awareness of how our distributed services interact.
What tools or technologies does your team use to support scalability, and why?
As part of this recent overhaul, we’ve introduced new languages, frameworks and philosophies into our tech stack, all with the goal of building software that can handle larger datasets while improving our end user’s experience. On the service layer, we’ve introduced Go in order to enable our move to a microservices application.
As a middleware, we’ve introduced GraphQL into the stack to handle resolving requests made from our web app to multiple services at once. And on the front end, we’ve revamped the UI and UX to gracefully handle the demand for accessing and manipulating large amounts of data. The amount of data flowing from server to client increases as we onboard larger enterprise customers, and it’s these technologies that allow us to scale to meet their needs.
