Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Optimizing FaaS Platforms for MCP-enabled Agentic Workflows

Varad Kulkarni, Vaibhav Jha, Nikhil Reddy, Yogesh Simmhan

TL;DR

The paper tackles the challenge of scalable deployment and state management for MCP-enabled agentic workflows on FaaS. It introduces FAME, a modular FaaS-based architecture that decomposes ReAct into Planner, Actor and Evaluator and orchestrates them with LangGraph and AWS Step Functions. It automates memory persistence via DynamoDB and optimizes MCP deployment with AWS Lambda wrappers, S3-based caching, and function fusion strategies. The experimental evaluation on research paper summarization and log analytics shows substantial improvements in latency, token efficiency, and cost, demonstrating the practicality of serverless hosting for complex multi-agent AI workflows.

Abstract

Agentic workflows that use autonomous AI Agents powered by Large Language Models (LLMs) and Model Context Protocol (MCP) servers is rapidly rising. This introduces challenges in scalable cloud deployment and state management. Traditional hosting on Virtual Machines (VMs) is resource-intensive and lacks elasticity. Functions-as-a-Service (FaaS) platforms offer modularity, autoscaling and cost efficiency but are inherently stateless. In this paper, we present the FAME, a FaaS-based architecture for orchestrating MCP-enabled agentic workflows. FAME decomposes agentic patterns such as ReAct into composable agents: Planner, Actor and Evaluator, that are each a FaaS function built using LangGraph and are orchestrated as a FaaS workflow. This enables modular composition as AWS Step Functions and avoids function timeouts seen for monolithic agentic workflows. To address context persistence across user requests in a conversation, FAME automates agent memory persistence and injection using DynamoDB. It also optimizes MCP server deployment through AWS Lambda wrappers, caches tool outputs in S3 and proposes function fusion strategies. We evaluate FAME on two representative applications, on research paper summarization and log analytics, under diverse memory and caching configurations. Results show up to 13x latency reduction, 88% fewer input tokens and 66% in cost savings, along with improved workflow completion rates. This demonstrates the viability of serverless platforms for hosting complex, multi-agent AI workflows at scale.

Optimizing FaaS Platforms for MCP-enabled Agentic Workflows

TL;DR

The paper tackles the challenge of scalable deployment and state management for MCP-enabled agentic workflows on FaaS. It introduces FAME, a modular FaaS-based architecture that decomposes ReAct into Planner, Actor and Evaluator and orchestrates them with LangGraph and AWS Step Functions. It automates memory persistence via DynamoDB and optimizes MCP deployment with AWS Lambda wrappers, S3-based caching, and function fusion strategies. The experimental evaluation on research paper summarization and log analytics shows substantial improvements in latency, token efficiency, and cost, demonstrating the practicality of serverless hosting for complex multi-agent AI workflows.

Abstract

Agentic workflows that use autonomous AI Agents powered by Large Language Models (LLMs) and Model Context Protocol (MCP) servers is rapidly rising. This introduces challenges in scalable cloud deployment and state management. Traditional hosting on Virtual Machines (VMs) is resource-intensive and lacks elasticity. Functions-as-a-Service (FaaS) platforms offer modularity, autoscaling and cost efficiency but are inherently stateless. In this paper, we present the FAME, a FaaS-based architecture for orchestrating MCP-enabled agentic workflows. FAME decomposes agentic patterns such as ReAct into composable agents: Planner, Actor and Evaluator, that are each a FaaS function built using LangGraph and are orchestrated as a FaaS workflow. This enables modular composition as AWS Step Functions and avoids function timeouts seen for monolithic agentic workflows. To address context persistence across user requests in a conversation, FAME automates agent memory persistence and injection using DynamoDB. It also optimizes MCP server deployment through AWS Lambda wrappers, caches tool outputs in S3 and proposes function fusion strategies. We evaluate FAME on two representative applications, on research paper summarization and log analytics, under diverse memory and caching configurations. Results show up to 13x latency reduction, 88% fewer input tokens and 66% in cost savings, along with improved workflow completion rates. This demonstrates the viability of serverless platforms for hosting complex, multi-agent AI workflows at scale.
Paper Structure (35 sections, 7 figures, 1 table)

This paper contains 35 sections, 7 figures, 1 table.

Table of Contents

  1. Introduction
  2. Background and Motivation
  3. Architecture of FAME
  4. Orchestrating Agentic Workflows on FaaS
  5. Durable Agent Memory with FaaS
  6. Deploying MCP Servers using FaaS
  7. Automated FaaS Wrappers
  8. Optimizations for FaaS MCP
  9. MCP Invocation Caching using S3
  10. S3-based File Handling
  11. Singleton vs. Consolidated MCP Servers
  12. Agentic Workflows and Configurations
  13. Agentic Applications
  14. Prompt Engineering
  15. Experimental Results
  16. ...and 20 more sections

Figures (7)

  • Figure 1: Agentic Workflow using ReAct Pattern
  • Figure 2: Architecture of FAME for a ReAct Workflow
  • Figure 3: Client (CM) and Agent Memory (AM) Injection, along with User Request (UR) into ReAct Workflow (P,A,E).
  • Figure 4: End to End Workflow Execution Latencies
  • Figure 5: Input and Output Tokens Consumed by LLM, per Workflow Invocation
  • ...and 2 more figures