Design and Evaluation of a Rack-Scale Disaggregated Memory Architecture For Data Centers

Puri, Amit; Jose, John; Tamarapalli, Venkatesh

doi:10.1109/hpcc-dss-smartcity-dependsys57074.2022.00060

Cited by 3 publications

(2 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The remote memory can be reserved in chunks of size, say 4MB to 16MB, whenever a node requests. DRackSim models distributed approach and uses another layer of an address mapping at compute nodes, as shown in Figure 5 for translating the local physical addresses to the remote physical address, as described in previous work [23]. This mapping is required to access allocated remote memory chunks, which differs from virtual address translation at TLBs.…”

Section: Remote Memory Organizationmentioning

confidence: 99%

“…A global manager handle conflicts during remote memory reservations to different nodes and acts as a load balancer for choosing a memory pool for allocation. As pointed out by [23], memory pools are bound to face contention in their queues when several compute nodes (with different memory access patterns) simultaneously access a remote memory pool. Memory pool selection should be done so that all pools face the same number of memory requests and avoid contention as much as possible.…”

Section: Global Memory Managermentioning

confidence: 99%

See 1 more Smart Citation

Design, Modeling, and Analysis of Memory Allocation Policies for Rack-Scale Memory Disaggregation

Puri

Jose

Tamarapalli

2023

Preprint

View full text Add to dashboard Cite

Memory disaggregation architecture is an emerging alternative to the traditional server architecture to address the issue of memory under-utilization in data centers. Memory disaggregation offers a large pool of remote memory connected via high-speed network fabric, thereby allows independent upgrade of memory components. Disaggregated design gives better adaptability of servers to the ever-increasing memory requirement from applications, such as HPC workload, video and big data analytics. However, disaggregrated memory architectures cause significantly large memory access latency due to the limited speed of interconnects currently used. In this paper, we propose a rack-scale disaggregated memory architecture and discuss the various software/hardware design aspects for such an architecture. We analyze the remote memory access latency in this architecture and observe that a significant part of the latency arises out of not only the interconnect, but also due to the contention at the queues of remote memory. To address this issue, we propose two-phase memory allocation policies for disaggregated memory environments that could significantly reduce the average memory access latency, compared to the conventional policies. We built a simulator that uses a trace-based front-end module and a cycle-accurate memory simulator to evaluate the memory access latency in a disaggregated memory system. We conduct experiments with different allocation policies across various benchmarks, representing applications with diverse memory access patterns. Our study shows encouraging results toward adoption of rack-scale memory disaggregation with acceptable memory latency, that could further be reduced with other system-level optimizations.

show abstract

Section: Remote Memory Organizationmentioning

confidence: 99%

Section: Global Memory Managermentioning

confidence: 99%