Why LambdaStore?

Minimizes Data Movement

Existing serverless platforms strictly separate execution and storage, which harms the performance of data-intenstive applications. LambdaStore co-locates storage and execution, which allows processing read and write operations almost instantaneously.

Provides Transactional Workflows

Conventional serverless platforms do not provide any transactional guarantees for data accesses.

Serverless workflows in LambdaStore are encapsulated in an ACID transaction, ensuring strong consistency and workflow atomicity. Writes of a job are only become visible once the associated workflow finished successfully. If a workflow fails, all changes to the data made by its jobs are rolled back.

This makes it significantly easier to avoid common concurrency errors. Developers do not have to implement their own locking on concurrency control mechanisms.

Ensures Atomic Job Execution

While existing serverless platforms might exeucte a function multiple times, LambdaStore executes functions exactly once. Because of this, developers do not need to write idempotent code and can even build non-deterministic programs if they want to.

Lightweight Job Isolation

LambdaStore leverages WebAssembly for quick and efficient isolation of jobs. A cold start takes less than one millisecond and allows your application to respond to client requests quickly.

Speeds up your Applications

The system is designed from scratch to be an efficient serverless and execution engine. LambdaStore is "cloud-native" by design and does not use any legacy code.

It outperforms existing platforms, such as OpenWhisk or OpenLambda, by an order of magnitude in our benchmarks.

Example Application

Coming soon...