EVENTGet 50% off your ticket to MongoDB.local NYC on May 2. Use code Web50! Learn more >

India's leading enterprise eCommerce platform cuts release cycles by 2X after migrating to MongoDB Atlas

StoreHippo slashes storage consumption by 50%, reduces latency by 30% and lowers DevOps overhead by 10% with MongoDB Atlas and Atlas Search

INDUSTRY

eCommerce and Retail
Computer Software/SaaS

PRODUCTS

MongoDB Atlas
Atlas Search

USE CASE

Catalog
Content Search
Payments

CUSTOMER SINCE

2021
INTRODUCTION

Serving thousands of medium and large enterprises across three continents

Launched in 2014, StoreHippo is a fully managed SaaS-based ecommerce platform. Built by Hippo Innovations, StoreHippo provides comprehensive turnkey solutions for B2C and B2B enterprises to build, scale, manage and market their online stores and multi-seller marketplaces. StoreHippo removes unnecessary IT burden from the shoulders of merchants, allowing them to focus on their core business.

StoreHippo has grown to become the leading Indian ecommerce SaaS platform focused on medium to large enterprises. It counts thousands of businesses as customers across more than 40 different industry sectors in 25+ Asia Pacific, European and Middle Eastern markets. The platform serves over 500 million secure API calls per month with less than 50 milliseconds average response times.

The company’s engineers originally built out its StoreHippo ecommerce platform with MongoDB community edition providing database services, and bolted-on external search engines to provide product catalog and content search.

To support the company’s rapidly scaling business and to increase the velocity of new feature releases, it migrated to the fully managed MongoDB Atlas platform and started offering Atlas Search as an alternative to Algolia and Azure Cognitive Search. As a result, the company has accelerated its speed in delivering new ecommerce services to market while further improving customer experience and reducing operational costs.

Beautiful, engaging, and high converting ecommerce storefronts using powerful search with autocomplete, built on the StoreHippo platform and MongoDB Atlas

Fig. 1: Beautiful, engaging, and high converting ecommerce storefronts using powerful search with autocomplete, built on the StoreHippo platform and MongoDB Atlas

THE CHALLENGE

Self-managing databases and bolting-on external search engines

Providing fast and engaging mobile-first experiences, all of the ecommerce merchant stores hosted on the StoreHippo platform are built as PWAs (Progressive Web Apps). Maximizing customizability and development velocity, StoreHippo’s ecommerce platform follows the MACH architectural pattern (Microservices, API-first, Cloud-native, Headless), providing a decoupled and scalable technology foundation.
StoreHippo’s MACH architecture with MongoDB powering the microservices’ data persistence layer

Fig. 2: StoreHippo’s MACH architecture with MongoDB powering the microservices’ data persistence layer

MongoDB was selected as the underlying database layer for StoreHippo because of its ease of integration with the Javascript-based MEAN Stack (MongoDB, Express, Angular, and Node). MongoDB’s flexible document data model allowed StoreHippo’s developers to handle the massive variability in product catalog and content attributes, along with associated metadata, media assets, and user reviews. Built on a distributed architecture, MongoDB provided the engineering teams with cloud-native resilience and scale-out to handle peak loads, especially over busy holiday seasons.

“We have used traditional relational databases such as MySQL in the past. The rigid schema of those databases inhibits how we can customize the product catalog for each customer. This customization is a key differentiator for StoreHippo over traditional monolithic ecommerce platforms. When you come to scale those databases, you typically have to re-engineer and denormalize your schema. With MongoDB, you have none of those limitations. As a result we can adapt and scale 3x faster.”

Rajiv Kumar Aggarwal, CEO and Founder, Hippo Innovations

MongoDB powers StoreHippo’s entire data persistence layer – from product catalog to order, inventory and customer management, through to payment processing. For security and performance isolation, each merchant is provisioned with their own single tenant MongoDB database and search engine.

The StoreHippo team initially opted to self-manage MongoDB on top of Microsoft Azure. However, self-managing MongoDB added operational overhead in running the systems, with the company’s engineers handling all patches, upgrades, scaling, backups, and performance optimizations.

Beyond the database, fast and relevant search is critical to help merchants’ customers quickly discover and buy products online. The StoreHippo platform offers customers a choice of using the Algolia or Azure Cognitive Search engines. However this makes the technology stack complex and costly. By bolting Algolia or Azure Cognitive Search onto their MongoDB database, the StoreHippo team has to grapple with:

  1. Writing their own synchronization mechanism via a set of custom hooks in the application code. This separate component pushes changes in the database to the search engine. Whenever the catalog’s schema changes, the developers have to remap the search indexes. With so many customers from so many different industry sectors, schema changes are constant, adding friction to onboarding new ecommerce sites and product offerings.
  2. Working with multiple query languages and drivers to query the database and search indexes. This slows down the pace of application development and impacts query latency. With Algolia and Azure Cognitive Search, search queries are first sent to the search engine, which returns a set of document identifiers. A follow-on query then has to be sent to the product catalog in MongoDB to return the source documents. This all slows down how quickly results are sent back to the user.
  3. Scaling and manual operational overhead as more stores are added to the StoreHippo platform.

Technology sprawl with three separate systems creating architectural complexity

Fig. 3: Technology sprawl with separate systems creating architectural complexity

THE SOLUTION

Migrating to fully-managed database services with integrated relevance-based search

To keep pace with the rapid scaling demands of the business, the company took the decision to migrate its self-managed MongoDB databases to fully-managed MongoDB Atlas.
“Our decision was driven by the operational efficiencies MongoDB Atlas unlocked. We can take advantage of the latest releases much faster, using new MongoDB features to build richer services for our customers. We have MongoDB engineers running the service using all of their established best practices, so my team can focus on the customer. It is now much easier for us to use advanced data placement policies like multi-region replication to push data closer to our customers in Europe and Australia, providing higher performance and an even better customer experience.”

Rajiv Kumar Aggarwal, CEO and Founder, Hippo Innovations

Before embarking on its migration to MongoDB Atlas, the StoreHippo team evaluated Azure CosmosDB, but quickly disqualified it because of a host of technical limitations:


  1. MongoDB Atlas offers 8x larger document sizes, providing much more flexibility in data modeling and storage.
  2. CosmosDB lacked many of the query and aggregation pipeline features StoreHippo’s developers rely on for the ecommerce service, which risked degrading application functionality for its customers.
  3. As CosmosDB is proprietary to Azure, the company would be locked-in to that platform.
“MongoDB Atlas gave us a superior fully managed service experience than CosmosDB. It also provided us with auto-scaling and multi-cloud flexibility. We are starting to move some of our tenants away from Azure onto Google Cloud to take advantage of that platform’s analytic features and optimized performance for data intensive applications like ours. With a few clicks in the Atlas UI, we can move customer data on-demand, with no downtime.”

Kriti Aggarwal, CMO and Co-Founder, Hippo Innovations

By moving to MongoDB Atlas, StoreHippo has also been able to take advantage of Atlas Search, offering it as a native search solution alongside the existing bolt-on Algolia and Azure Cognitive Search services.

By using Atlas Search the StoreHippo engineering team has streamlined its technology estate. Atlas Search is built on the industry standard Apache Lucene library. Therefore developers are able to deliver the key capabilities demanded for fast and relevant catalog and content search:

  • Autocomplete automatically surfaces suggested products as a customer starts typing in the search bar. This helps them find products faster.
  • Lightning-fast facets and counts helps customers efficiently navigate categorized search results.
  • Relevance-based tuning with scoring allows StoreHippo’s merchants to control the order of search results returned to the customer. This allows them to rank preferred products and promotions at the top of the results set, increasing clickthrough rates and conversions.
  • Over 40 multilingual language analysers provide customers with a native search experience, no matter which country they are in.

StoreHippo’s developers are more productive as they now work with a unified API and single MongoDB Node.js driver across both database and search operations. This integrated experience eliminates them having to context switch between different MongoDB and custom search engine query languages as they code, and removes unnecessary build dependencies.

Atlas Search is integrated into the MongoDB aggregation pipeline, therefore developers can combine multiple aggregation stages and operators to build more sophisticated queries. For example, the product catalog can store multiple descriptions for each product in different languages. Fast and dynamic filtering in the MongoDB query API allows the correct description to be returned, based on the customer’s language-specific search term.

Improving data consistency, Atlas automatically synchronizes all changes made to the catalog database with the Atlas search index. This avoids the engineering team having to manually build and maintain their own fragile custom replication mechanism and transformation logic.

Engineering cycles spent on building and maintaining the previously separate database, search engine and sync mechanism are now invested back into building new features and applications for the business.

Dramatic architectural simplification with fully integrated database, sync, and search in MongoDB Atlas

Fig. 4: Dramatic architectural simplification with fully integrated database, sync, and search in MongoDB Atlas

THE RESULTS

2X Faster release cycles with 50% reduction in storage overhead and 30% lower latency

Migrating from the self-managed community edition of MongoDB to MongoDB Atlas has reduced StoreHippo’s operational overhead and created a more efficient data infrastructure.
“I’d estimate running in MongoDB Atlas saves around 10% of our total engineering cycles, which we can now reinvest in innovation. Also using Atlas’ native disk compression saves 50% storage capacity and costs.”

Rajiv Kumar Aggarwal, CEO and Founder, Hippo Innovations

Moving its native search functionality to Atlas Search has also unlocked benefits in time to market and user experience:

  • 2x faster for developers to roll out new features because they now write code in the unified MongoDB query API. This avoids both the cognitive and testing overhead of having to switch between multiple systems.
  • 30% lower query latency. MongoDB Atlas brings previously separate queries that were run independently against the bolt-on search engine and database, into a single application call. This enables retrieval of both the identifiers and matching documents in one round trip to the data layer. Back in 2008, Amazon published some famous research quantifying each 100ms of additional latency cost 1% in sales. Since then, customers have become more demanding. Any reduction in latency has an immediate impact on retailers' clickthrough rates and conversions.
  • Fresher, more up to date product data as sync errors between the database and search index have been eliminated, further improving customer experience.
Now that StoreHippo is running on MongoDB Atlas, the company’s engineering team is starting to explore other data services available to them in the platform. These services will help them further improve merchant and customer experiences:
  • The MongoDB Atlas Data Federation will allow StoreHippo to archive older data out of tenants’ database into low cost cloud object storage. By keeping the data queryable, developers and data scientists can combine live operational data in the database with archived data in a Federated Database Instance – including data generated outside of MongoDB – to power real time product recommendations and better predict buyer intent.
  • The effectiveness of promotional campaigns can be easily exposed to merchants in rich visualizations powered by MongoDB Charts.
  • The MongoDB Data API will allow StoreHippo developers to query the database directly from the client browser over HTTPS, enabling faster delivery of new frontend features.
If you are interested in working on the cutting edge of retail technology, take a look at the open job postings available at StoreHippo.

NEXT STEPS

Get started with MongoDB Atlas today

If you are bolting external search engines onto your databases today, we can help you simplify your application estate. Through the experiences gained by working with customers who have migrated, we have put together a repeatable 5-step methodology to replacing bolt-on search engines with MongoDB Atlas Search.

What will your story be?

MongoDB will help you find the best solution.