Apiary: A DBMS-Backed Transactional Function-as-a-Service Framework

  title={Apiary: A DBMS-Backed Transactional Function-as-a-Service Framework},
  author={Peter Kraft and Qian Li and Kostis Kaffes and Athinagoras Skiadopoulos and Deeptaanshu Kumar and Danny Hyun Bum Cho and Jason J. Li and Robert Redmond and Nathan Weckwerth and Brian Xia and Peter D. Bailis and Michael J. Cafarella and Goetz Graefe and Jeremy Kepner and Christos Kozyrakis and Michael Stonebraker and Lalith Suresh and Xiangyao Yu and Matei A. Zaharia},
Developers are increasingly using function-as-a-service (FaaS) platforms for data-centric applications that primarily perform lowlatency and transactional operations on data, such as for microservices or web serving workloads. Unfortunately, existing and recently proposed FaaS platforms support these applications poorly because they separate application logic, executed in cloud functions, from data management, done in interactive transactions accessing remote storage. This separation harms… 



Distributed transactions on serverless stateful functions

A programming model and implementation for transaction orchestration of stateful serverless functions and design and implement the model on Apache Flink StateFun, which supports serializable distributed transactions with two-phase commit, as well as relaxed transactional guarantees with Sagas.

Transactional Causal Consistency for Serverless Computing

This work considers the setting of serverless Function-as-a-Service (FaaS) platforms, where storage services are disaggregated from the machines that support function execution, and presents protocols for Multisite Transactional Causal Consistency (MTCC), which provides causal consistency for all I/Os within a given transaction even if it runs across multiple physical sites.

Starling: A Scalable Query Engine on Cloud Functions

Starling is a query execution engine built on cloud function services that employs a number of techniques to mitigate challenges, providing interactive query latency at a lower total cost than provisioned systems with low-to-moderate utilization.

Cloudburst: Stateful Functions-as-a-Service

Empirical results show that Cloudburst makes stateful functions practical, reducing the state-management overheads of current FaaS platforms by orders of magnitude while also improving the state of the art in serverless consistency.

Narrowing the Gap Between Serverless and its State with Storage Functions

Shredder, a low-latency multi-tenant cloud store that allows small units of computation to be performed directly within storage nodes, achieves a 14% to 78% speedup against conventional remote storage when fetching items with just one to three data dependencies between them and is effective in accelerating data-intensive applications.

Faasm: Lightweight Isolation for Efficient Stateful Serverless Computing

Faaslets, a new isolation abstraction for serverless big data computing, is introduced and it is shown that, when training a machine learning model, it achieves a 2x speed-up with 10x less memory; for serving machine learning inference, Faasm doubles the throughput and reduces tail latency by 90%.

Scalable distributed transactions across heterogeneous stores

This work proposes an approach that enables multi-item transactions with snapshot isolation across multiple heterogeneous data stores using only a minimal set of commonly implemented features such as single item consistency, conditional updates, and the ability to store additional meta-data.

CPU Sharing Techniques for Performance Isolation in Multitenant Relational Database-as-a-Service

This paper shows that traditional CPU sharing abstractions and algorithms are inadequate to support several key new requirements that arise in DaaS: (a) absolute and fine-grained CPU reservations without static allocation; (b) support elasticity by dynamically adapting to bursty resource demands; and (c) enable the DAAS provider to suitably tradeoff revenue with fairness.

LambdaObjects: Re-aggregating storage and execution for cloud computing

The early results demonstrate that this architecture provides lower latencies, while sacrificing some elasticity offered by conventional serverless systems, and LAMBDAOBJECTS provide strong consistency and enable efficient caching mechanisms and fault-tolerance with low overhead.

Software-Defined Data Protection: Low Overhead Policy Compliance at the Storage Layer is Within Reach!

The approach is simple, yet powerful: decoupling often changing policies from request-level enforcement allows distributed smart storage nodes to implement the latter at line-rate, both for company-specific rules and regulations, such as GDPR or CCPA.