That DOS Won’t Hunt: Shutterfly CTO Talks Going Cloud

Host’s Note: This episode of the That DOS Won't Hunt podcast ran into an audio issue during recording, so the content has been transcribed in part below. An incorrect recording software setting caused the issue and will not be a problem in the future.

That is the theme of the podcast after all -- identifying tech problems and exploring options to address them.

Listen to "That DOS Won't Hunt" on Spreaker.

Introduction

Imagine your organization runs a substantial set of well-known brands with a robust ecommerce platform that has been operating from an on-prem datacenter. The day may come when migration to cloud-based resources becomes not only an option but a strategic opportunity. Even then, it might not be a snap decision or a process that happens overnight.

Why make such a move? What considerations does the organization have to weigh during such an undertaking? Was it worth the effort?

Today I will be speaking with Shutterfly’s CTO Moudy Elbayadi about the company’s migration of its ecommerce platform to MongoDB Atlas on AWS. I’m your host João-Pierre Ruth, and welcome to That DOS Won’t Hunt.

So, let’s get started. Moudy, thank you so much for being with us today. We appreciate you making the time.

Thank you and it’s good to be here with you.

Give us the scope of Shutterfly’s operation and brands. I’m sure our listeners will be familiar with them but give them a bit of an understanding of the needs and demands that you were facing and how the ecommerce platform supported the growth of the brands.

Shutterfly, as a way of background, began its mission almost 20 years ago and the mission initially was around photo sharing, photo gathering, and we have moved into the photo personalization space. And in that timeframe, we have hosted over 100+ billion photos served over 20 million customers and it’s about capturing those important moments in people’s lives. Since becoming a household brand, we have leveraged those special moments with Snapfish and the most recent part of the brand, Spoonflower, which is all based on custom design. For us at Shutterfly, we’re very, very focused on really expanding that and bringing more of the ability to personalize and bringing that to many more assets and the things that you care about.

From a technology perspective, we have a mission on “How do we really delight customers?” in all that we do. And from a technology perspective, when we think about what does delight mean, it really meant that we needed to spend more time focusing on improving the customer experience and spending less time on infrastructure and the back office. And we’ll speak a little more about what that looks like and what it looked like for us. Ultimately, a big part of getting that focus back was really centered around how do leverage the power of the public cloud and the power of hosted services.

So that was really what prompted discussions to make that shift of the ecommerce platform to the cloud? Is that really what got things rolling?

A big part of -- for anyone who’s managed local infrastructure on-premise, on-premise data center infrastructure, one of the most precious resources we have is management’s attention and organizational attention. Over the time we managed these services, a little bit of how much attention goes to managing, maintaining, monitoring, developing, supporting, securing the infrastructure -- how much time and attention you give to focus on “What are the kinds of services, what are products that will delight customers?” And what we found was -- obviously there are aspects of “Is it cheaper to run in public cloud? Is it more efficient?”

Obviously, there are a lot of different dimensions to migrating to the cloud and leveraging services like MongoDB Atlas. Really, one of the key kernels that I continue to learn is that the ability to refocus your organization’s energy on what will delight the customers more because when I’m running on this particular storage frame or this particular network or not, you know the end customer doesn’t really care about that. What they care about is “Is this experience easy? Is it intuitive? Are you bringing in machine learning in the background to help me create this photobook? To help me pick the photos that I want?”

That is ultimately the delight from all those things that powers that experience versus if you are stuck in a more traditional datacenter model. You get to spend a lot more time just managing all of those things.

Did you have a checklist of considerations, concerns that really needed to be addressed as you were taking that look into migration? Be it costs, security. Were there any functional roadblocks that at first blush might have seemed insurmountable to make that transition -- that you needed some time soul-searching about “How is this going to work?”

If you’re talking about just the migration from an on-premise environment to a cloud-based environment -- I think there’s a number of dimensions and some that we don’t talk enough about is there is the people dimension, there is the culture dimensions, there is the business stakeholder and envisioning dimension, and then there’s the actual work, which is “We’re going to shift from this technology and this design pattern, architectural pattern to this design.”

I think the challenge in many ways is we have to be able to address all of those dimensions. If you ignore the people dimension, you can have a situation where you have loyal employees focused on delivering a set of services and capabilities and skills, and as that shifts, what is both the vision and also creating a path for making sure that your team is able to go along with you on this journey.

There’s that aspect of people and team readiness, culture readiness. On the other hand, some of this technology and paradigms will not work in the public cloud. One of the worst things you can do is continue -- I call it the horseless carriage syndrome -- to operate or look at the public cloud as if it were your datacenter. It really requires a mindset and a paradigm shift; you don’t want to run it the same way you ran a datacenter.

When cars were first invented, the first person was like “Ah, I see it. It looks like a horseless carriage,” not realizing that a car is different thing than a horseless carriage. There is a lot more that can happen with a car that you can do. It opens up a lot more possibilities with an automobile than a horse drawn carriage. In many ways, there’s almost like there’s a paradigm shift of “This is not just a datacenter with servers.” It’s really the code that powers and instruments and orchestrates everything. It’s no longer having these dedicated sets of servers and systems. So, it’s very much a paradigm shift.