SC19 Proceedings

The International Conference for High Performance Computing, Networking, Storage, and Analysis

Abstract: Emerging Non-Volatile Memories have byteaddressability and low latency, close to the latency of main memory, together with the non-volatility of storage devices. Similarly, recently emerging interconnect fabrics, such as Gen- Z, provide high bandwidth, together with exceptionally low latency. These concurrently emerging technologies are making possible new system architectures in the data centers including systems with Fabric-Attached Memories (FAMs). FAMs can serve to create scalable, high-bandwidth, distributed, shared, byteaddressable, and non-volatile memory pools at a rack scale, opening up new usage models and opportunities.

Based on these attractive properties, in this paper we propose FAM-aware, checkpoint-based, post-copy live migration mechanism to improve the performance of migration. We have implemented our prototype with a Linux open source checkpoint tool, CRIU (Checkpoint/Restore In Userspace). According to our evaluation results, compared to the existing solution, our FAMaware post-copy can improve at least 15% the total migration time, at least 33% the busy time, and can let the migrated application perform at least 12% better during migration.






Back to MCHPC’19: Workshop on Memory Centric High Performance Computing Archive Listing


Back to Full Workshop Archive Listing