Today’s guest blog comes from Rafael Sanches, engineer at Allthecooks, a social media platform for people who love cooking. Allthecooks is available on Android, iPhone, Windows Phone, Google Glass and Android Wear, and is a top recipe app on Google Play.



At Allthecooks, we’re connecting passionate chefs with casual and first-time cooks on every major mobile device including Android phones, tablets, watches and Google Glass. People use our app to find, rate and comment on dishes they can cook themselves, or post their own ideas complete with directions, ingredients, servings and nutrition information. We have chefs with tens of thousands of followers on Allthecooks, meaning that whenever they make an update we have to process hundreds of thousands of simultaneous API requests to feed that information to timelines of all their followers.



Creating a successful platform isn’t just about speed, it’s about scalability, too. Google Cloud Platform played a key role in helping us grow without worrying about our architecture. We launched in December 2012 with just three part-time engineers and have never taken funding, so building our own infrastructure was out of the question. Since launching, we’ve grown to over 12 million users with a million monthly active users. Our application now sees millions of interactions daily that run through Google App Engine and Google Cloud Datastore.



As our user base has grown, we’ve begun migrating the biggest pieces of our backend processing architecture from App Engine onto Google Compute Engine. This will allow us to operate at even higher performance levels by using the cheaper CPU and caching more data in the large RAM configurations supported by Compute Engine instances. We also plan to use Google BigQuery soon to make the process of finding the perfect recipe even easier.



Cost was a big concern when we were burning through our savings to build Allthecooks, but it wasn’t as important as making everything run as fast as possible. When a request hits, the response needs to be immediate. That’s why we built our recommendation engine on Compute Engine -- that lets us run entirely on RAM, so users get the results they need as soon as they need them. When you’ve got 60 seconds to sear a fish just right, you can’t be caught waiting on a laggy server request to see the next step in a recipe. We never want our app latency to stand between our users and a great meal.



There are plenty of cooking apps out there, but none with the same level of social interaction as Allthecooks, which is one of the biggest reasons our users spend so much time on the app. We let users ask questions and receive answers on every recipe, and upload pictures of their own dishes. Our recommendation engine in particular plays a pivotal role in making the Allthecooks experience so useful on mobile devices. Google Glass users love that they can find and follow a recipe instructions hands-free or tilt their heads up to see the ingredient list. They can even record their own recipes using voice, photos and video and send it to their Allthecooks account. All of this requires a reliable infrastructure.



We don’t have a system administrator, so we need a system that’s reliable, manageable and scalable while requiring minimal oversight. Cloud Platform gives us confidence that the app is stable, and it can even be left alone for a week without requiring maintenance. Google’s support team has also given us peace of mind. They’re always quick to respond, and they have provided great services the few times we needed help.



It gives us great pride that Allthecooks has helped helped millions of people live healthier, and discover new foods and products they love. We believe in launching early and listening to what our customers want. Cloud Platform is a crucial component to our success to date and strategic for our future growth.



-Posted by Rafael Sanches, co-founder and engineer of Allthecooks