Real Estate

Rent the Runway embraces a code-first, dynamic data model approach to scale warehouse operations

Rent the Runway

Rent the Runway

Founded in ny town in 2009, Rent the Runway is disrupting the trillion-dollar garment industry and provoking ladies with its vision for a lot of joyful, financially savvy, and property thanks to feeling their best each day. Through Rent the Runway, customers will rent, buy, or subscribe to designer attire, accessories, and residential decoration from over 750 whole partners. whereas Rent the Runway aims for a seamless method for its customers, the back-end technology powering this one-of-a-kind provision operation is complicated.

The company has 2 distribution facilities — one in New Jersey and one in TX — that receive, clean, and repair things before fleetly packing and shipping them out once more. This involves a big investment in AI, radio ID tags, and robots to a lot of with efficiency type, clean, and ship the clothes.

Ahead of his presentation at MongoDB’s annual developer conference, Rent the Runway’s Director of Engineering, electro-acoustic transducer Liberant, mentioned the company’s knowledge strategy and code-first approach and the way that factors into the company’s end-user expertise.


MongoDB: are you able to describe what goes on covert at your distribution facilities once a garment is came by a customer?

ML: First, each garment that came to our warehouses should be cleansed and sorted before it’s obtainable to rent once more, therefore our goal is to alter the movement of products wherever doable to contour this method.

Worn clothes are initially ushered onto a conveyer before they are going through an associate degree setup. Why associate degree X-ray machine? Quite ofttimes, our members can accidentally leave things within the pockets of their rented clothes (lipstick could be a frequent offender) that we’d like to catch before they are going into a washer.

After it passes through the X-ray — hopefully makeup-free — our software package kinds things into one among 20+ completely different bins for cleanup. Adding to the challenge is that the proven fact that these clothes need completely different cleanup ways, which means that the robotic arms managing the sorting method have to be compelled to perceive that bins correlate to that cleanup methodology. These solutions ar vital to scaling our operations, and success comes right down to arming our development groups with tools that facilitate them to add up the huge amounts of knowledge generated.


MongoDB: however will this huge quantity {of knowledge|of knowledge|of information} impact your development groups and data model?

ML: Given our tight timeline, employing a {relational knowledgebase|electronic database|on-line database|computer database|electronic information service} was a no-go because of the direct necessities you would like to account for in terms of planning schema properly and therefore the corresponding data model. we have a tendency to knew there would be an outsized learning curve and that we would have to be compelled to systematically reiterate thereon knowledge model, that entails dynamic your schema for each iteration once employing a computer database.

This means an information administrator needs to coordinate with a development team to feature columns in, and therefore the developers then have to be compelled to return and update their code to match the corresponding knowledge. It’s a singular challenge for our engineers as a result of we’re operating with a spread of various clothes and we’re attempting to create a software package that may clean them properly.

MongoDB: however did your information selection impact the outcome?

ML: mistreatment MongoDB’s document knowledge model was key to reducing our developers’ psychological feature hundreds. It doesn’t need the direct work of planning our schemas and it permits the US to stay to a code-first approach.

As object minded software package developers, we are able to create by mental act the planet around US as objects simply. as an example, a automotive is associate degree object with bound attributes associate degreed a garment is an object with bound attributes. once that garment contains a specific cleanup methodology hooked up thereto, it’s simply another attribute. The document model permits the US to deploy, go live, and add attributes presently with little or no coordination between information directors. this suggests our developers write code precisely however it seems in their heads, rather than having to normalize the info into multiple tables.

MongoDB: will this knowledge strategy stretch across the company?

ML: Since connexion Rent the Runway in 2019, I even have centered on processing our variable star vision for building applications, {including|as we have a tendency toll as|together with} however we designer our systems. we’ve been able to separate every business domain mistreatment microservices backend by MongoDB Atlas, Kotlin, and Spring Boot, to produce a contemporary technical school stack.

During the pandemic, we have a tendency to centered on increasing potency through increased automation whereas arriving and departing volume was lower. Minimizing our arriving time interval isn’t solely sensible for our business, it helps US unlock inventory quicker therefore our customers will rent or expire. once a garment is picked up by a robotic arm, the arm scans the garment’s RFID tag and determines what bin it has to be sorted into. The robotic arm conjointly tells US once a bin is full and wishes to be stirred to a washer, once it’s empty and every one of these alternative types of knowledge points that ar helpful. something knowledge-related conjointly gets traced to our data deposit system.

Any time period in our warehouse creates a delay for our customers, therefore we have a tendency to proactively took measures to de-risk this by implementing triggers — that modify US to execute application and information logic mechanically — either in response to events or on a predefined schedule. mistreatment Realm Triggers inside Atlas to pipe knowledge to our knowledge warehouse is basically a no-code resolution that helped to additional de-risk our entire system, permitting US to extract the worth of this knowledge for a future statement of warehouse workloads.

MongoDB: What were the advantages of employing a multi-cloud information service?

ML: mistreatment Atlas, we have a tendency to reached our time to promote for the warehouse automation rollout in 0.5 the time of our bequest technical school stack. we have a tendency to achieved this strictly through this code-first, dynamic knowledge modeling approach and were able to improve the potency of our arriving sortation dramatically. At the tip of the day, this plays a key role in rising inventory accessibility and, therefore, making a superior client expertise.

Read More

Leave a Response