Caching Techniques for Security Metadata in Integrity-Protected Fabric-Attached Memories

Abstract

The constant need for larger memories and the diversity of workloads have driving the system vendorsaway from the conventional processor-centric architecture into a memory-centric architecture. Memorycentric architecture, allows multiple computing nodes to connect to a huge shared memory pool andaccess it directly. To improve the performance, each node uses a small local memory to cache the data.These architectures introduces several problems when memory encryption and integrity verification areimplemented. For instance, using a single integrity tree to protect both memories can introduce unnecessaryoverheads. Therefore, we propose Split-Tree, which implements two separate integrity tree for each memory.Later, we analyze the system performance, and the security metadata caches behavior when separate treesare used. We use the gathered insights to improve the security metadata caching for the separate trees andultimately improve the system performance.

Keywords

• fabric-attached memory
• secure memory
• memory-centric
• encrypted memory
• integrity tree