Optimization and Scaling

This page is for MSA Search, version 2.6.0 (Latest) (default). For other versions, use one of these documentation indexes: - 2.6.0 (Latest) (default): https://docs.nvidia.com/nim/bionemo/msa-search/2.6.0/llms.txt - 2.5.0: https://docs.nvidia.com/nim/bionemo/msa-search/2.5.0/llms.txt - 2.4.0: https://docs.nvidia.com/nim/bionemo/msa-search/2.4.0/llms.txt - 2.3.0: https://docs.nvidia.com/nim/bionemo/msa-search/2.3.0/llms.txt - 2.2.0: https://docs.nvidia.com/nim/bionemo/msa-search/2.2.0/llms.txt - 2.1.0: https://docs.nvidia.com/nim/bionemo/msa-search/2.1.0/llms.txt - 2.0.0: https://docs.nvidia.com/nim/bionemo/msa-search/2.0.0/llms.txt - 1.0.0: https://docs.nvidia.com/nim/bionemo/msa-search/1.0.0/llms.txt

Starting with version 2.0.0, MSA NIM requires less configuration for optimal performance. The GPU Server is enabled by default with pre-indexed ColabFold databases, so you do not need to prepare databases manually or tune complex configuration. For most users, the default settings provide good performance.

This section focuses on scaling strategies for production deployments that require high throughput or need to handle variable workloads.

Scaling and Load Balancing

With version 2.0.0, the MSA Search NIM no longer internally manages GPU allocation across multiple GPUs in a node. Instead, the NIM is designed to work with external load-balancers and orchestration platforms that can scale NIM instances up and down based on demand. Examples include NVIDIA Cloud Functions and container orchestration platforms that support scalable, distributed deployments of containerized applications.

The recommended approach is to allocate the minimum number of GPUs to each individual NIM instance and rely on horizontal scaling through orchestration. This allows for:

• Better resource utilization across your infrastructure
• Dynamic scaling based on actual workload demands
• Improved fault tolerance through instance redundancy
• Simplified deployment and maintenance