LinkBench is a database benchmark developed to evaluate database performance for workloads similar to those of Facebook's production MySQL deployment. LinkBench is highly configurable and extensible. It can be reconfigured to simulate a variety of workloads and plugins can be written for benchmarking additional database systems.
Background
One way of modeling social network data is as a social graph, where entities or nodes such as people, posts, comments and pages are connected by links which model different relationships between the nodes. Different types of links can represent friendship between two users, a user liking another object, ownership of a post, or any relationship you like. These nodes and links carry metadata such as their type, timestamps and version numbers, along with arbitrary payload data.
Facebook represents much of its data in this way, with the data stored in MySQL databases. The goal of LinkBench is to emulate the social graph database workload and provide a realistic benchmark for database performance on social workloads. LinkBench's data model is based on the social graph, and LinkBench has the ability to generate a large synthetic social graph with key properties similar to the real graph. The workload of database operations is based on Facebook's production workload, and is also generated in such a way that key properties of the workload match the production workload.
LinkBench Architecture
++====================================++ || LinkBench Driver || ++====================================++ || +---------------------------+ || || | Graph | Workload | || Open connections +=======+ || | Generator | Generator | || /------------------> | Graph | || +---------------------------+ || /-------------------> | Store | || | |<======---------------------> | Shard | || | Graph Store Adapter |<======---------------------> | | || | (e.g. MySQL adapter) |<======---------------------> | e.g. | || +---------------------------+ || \-------------------> | MySQL | || || \------------------> | Server| || ~~~~~~~~~~~~ ~~~~~~~~~~~~ || +=======+ || ~~~~~~~~~~~~ ~~~~~~~~~~~~ || || ~~~~~~~~~~~~ ~~~~~~~~~~~~ || || ~~~~~~~~~~~~ ~~~~~~~~~~~~ || || Requester Threads || ++====================================++
The main software component of LinkBench is the driver, which acts as the client to the database being benchmarked. LinkBench is designed to support benchmarking of any database system that can support all of the require graph operations through a Graph Store Adapter.
The LinkBench benchmark typically proceeds in two phases.
The first is the load phase, where an initial graph is generated using the graph generator and loaded into the graph store in bulk. On a large benchmark run, this graph might have a billion nodes, and occupy over a terabyte on disk. The generated graph is designed to have similar properties to the Facebook social graph. For example, the number of links out from each node follows a power-law distribution, where most nodes have at most a few links, but a few nodes have many more links.
The second is the request phase, where the actual benchmarking occurs. In the request phase, the benchmark driver spawns many request threads, which make concurrent requests to the database. The workload generator is used by each request thread to generate a series of database operations that mimics the Facebook production workload in many aspects. For example, the mix of different varieties of read and write operations is the same, and the access patterns create a similar pattern of hot (frequently access) and cold nodes in the graph. At the end of the request phase LinkBench will report a range of statistics such as latency and throughput.