Design Considerations in Yelp

Introduction

We discussed the design, components, and workflow of our system in the previous lesson, and a number of interesting questions need answering. For example, what approach do we use to find the places, or how do we rank places based on their popularity? This lesson answers such important design concerns. The table below summarizes the goals of this lesson.

Searching

From Google maps, we were able to connect segments and meet the scalability challenge to process a large graph efficiently. The graph of the world had numerous nodes and vertices, and traversing them was time-consuming. So, we divided the whole world into smaller segments/subgraphs to process and query them in parallel. The segmentation helps in improving the scalability of the system.

Each segment will be of size $5 \times 5 \ miles$ and have a list of places in it that exist within it. We can search only a few segments to locate close destinations. To identify nearby sites, we can use a given location and radius to find all nearby segments.

We can store all places in a table and uniquely identify a segment by having a segment_ID. We can index each segment in the database (see next section for details). Now we have limited the number of segments we need to search so the query will be optimized and return results quickly.

Improving data fetching

We will use a key-value store for quick access to the places in the segments. The key will be the segment_ID while value will have the list of places in that segment. Let’s estimate how much space we will need to store the indexes.

We can usually store a few MBs in the value of the key-value store. If we assume that each segment has 500 places then it will take up $500 \times 1296 \ Bytes = 0.648MBs$ and we can easily ...