Model inference systems are essential for implementing end-to-end data analytics pipelines that deliver the benefits of machine learning models to users. Existing cloud-based model inference systems are costly, not easy to scale, and must be trusted in handling the models and user request data. Serverless computing presents a new opportunity, as it provides elasticity and fine-grained npricing. Our goal is to design a serverless model inference system that protects models and user request data from untrusted cloud providers. It offers high performance and low cost, while requiring no intrusive changes to the current serverless platforms. To realize our goal, we leverage trusted hardware. We identify and address three challenges in using trusted hardware for serverless model inference. These challenges arise from the high-level abstraction of serverless computing, the performance overhead of trusted hardware, and the characteristics of model inference workloads. We present SeSeMI, a secure, efficient, and cost-effective serverless model inference system. It adds three novel features non-intrusively to the existing serverless infrastructure and nothing else. The first feature is a key service that establishes secure channels between the user and the serverless instances, which also provides access control to models and users' data. The second is an enclave runtime that allows one enclave to process multiple concurrent requests. The final feature is a model packer that allows multiple models to be executed by one serverless instance. We build SeSeMI on top of Apache Open Whisk, and conduct extensive experiments with three popular machine learning models. The results show that SeSeMI achieves low latency and low cost at scale for realistic workloads.

Modern deep learning models capture the semantics of complex data by transforming them into high-dimensional embedding vectors. Emerging applications, such as retrieval-augmented generation, use approximate nearest neighbor (ANN) search in the embedding vector space to find similar data. Existing vector databases provide indexes for efficient ANN searches, with graph-based indexes being the most popular due to their low latency and high recall in real-world high-dimensional datasets. However, these indexes are costly to build, suffer from significant contention under concurrent read-write workloads, and scale poorly to multiple servers. Our goal is to build a vector database that achieves high throughput and high recall under concurrent read-write workloads. To this end, we first propose an ANN index with an explicit two-stage design combining a fast filter stage with highly compressed vectors and a refine stage to ensure recall, and we devise a novel lightweight machine learning technique to fine-tune the index parameters. We introduce an early termination check to dynamically adapt the search process for each query. Next, we add support for writes while maintaining search performance by decoupling the management of the learned parameters. Finally, we design HAKES, a distributed vector database that serves the new index in a disaggregated architecture. We evaluate our index and system against 12 state-of-the-art indexes and three distributed vector databases, using high-dimensional embedding datasets generated by deep learning models. The experimental results show that our index outperforms index baselines in the high recall region and under concurrent read-write workloads. Furthermore, HAKES is scalable and achieves up to 16x higher throughputs than the baselines.

Machine learning (ML) is an important part of modern data science applications. Data scientists today have to manage the end-to-end ML life cycle that includes both model training and model serving, the latter of which is essential, as it makes their works available to end-users. Systems of model serving require high performance, low cost, and ease of management. Cloud providers are already offering model serving choices, including managed services and self-rented servers. Recently, serverless computing, whose advantages include high elasticity and a fine-grained cost model, brings another option for model serving.