Advances in memory technologies ( e . g ., HBM, DRAM, NVM) and interconnects ( e . g ., CXL) have significantly enhanced the flexibility of utilizing memory resources in modern computer systems. As this trend continues, memory resources are poised to become fully composable in the near future. This increasing flexibility also accelerates the demand for effective tiered memory systems capable of performing well across diverse scenarios and workloads. However, the effectiveness of existing tiered memory systems heavily depends on system configuration (e.g., the ratio of fast tier to capacity tier), memory access patterns, and page sizes (base vs. huge). Their reliance on simple heuristics and static thresholds for detecting page hotness, and limited consideration of page sizes, results in suboptimal (often pathological) page placement decisions. To build a robust and effective system, tiered memory management must holistically account for overall memory access patterns in conjunction with the tiering environment. We present Memtis , a tiered memory system that adopts informed decision-making for page placement and page size determination. Memtis leverages access distribution of allocated pages to optimally approximate the hot data set to the fast tier capacity. Moreover, Memtis dynamically determines the page size that allows applications to use huge pages while avoiding their drawbacks by detecting inefficient use of fast tier memory and splintering them if necessary. To further enhance its practicality across diverse scenarios, Memtis effectively supports the dynamic allocation of fast tier memory if there is a specific performance requirement, such as a target hit ratio. Our evaluation shows that Memtis outperforms existing tiered memory systems under various workloads and tiering configurations by up to 169.0%, showing its robustness.

As data generation has been on an upward trend, storing vast volumes of data cost-effectively as well as efficiently accessing them is paramount. At the same time, today's storage landscape continues to diversify, from high-bandwidth storage devices such as NVMe SSDs to low-latency non-volatile memory (e.g., Intel Optane DCPMM). These heterogeneous storage devices have the potential to deliver high performance in terms of bandwidth and latency with cost efficiency, while achieving the performance and cost targets together still remains a challenging problem. We provide our solution, Prism, a novel key-value store that utilizes modern heterogeneous storage devices. Prism uses heterogeneous storage devices synergistically to harness the advantages of each storage device while suppressing their downsides. We devise new techniques to balance the latency-bandwidth tradeoff when reading from SSD. For ensuring multicore scalability and crash consistency of data across heterogeneous storage media, Prism proposes cross-storage concurrency control and cross-storage crash consistency protocols. Our evaluation shows that Prism outperforms state-of-the-art key-value stores by up to 13.1× with significantly lower tail latency.

The evergrowing memory demand fueled by datacenter workloads is the driving force behind new memory technology innovations (e.g., NVM, CXL). Tiered memory is a promising solution which harnesses such multiple memory types with varying capacity, latency, and cost characteristics in an effort to reduce server hardware costs while fulfilling memory demand. Prior works on memory tiering make suboptimal (often pathological) page placement decisions because they rely on various heuristics and static thresholds without considering overall memory access distribution. Also, deciding the appropriate page size for an application is difficult as huge pages are not always beneficial as a result of skewed accesses within them. We present Memtis, a tiered memory system that adopts informed decision-making for page placement and page size determination. Memtis leverages access distribution of allocated pages to optimally approximate the hot data set to the fast tier capacity. Moreover, Memtis dynamically determines the page size that allows applications to use huge pages while avoiding their drawbacks by detecting inefficient use of fast tier memory and splintering them if necessary. Our evaluation shows that Memtis outperforms state-of-the-art tiering systems by up to 169.0% and their best by up to 33.6%.

RDMA-capable networks are gaining traction with datacenter deployments due to their high throughput, low latency, CPU efficiency, and advanced features, such as remote memory operations. However, efficiently utilizing RDMA capability in a common setting of high fan-in, fan-out asymmetric network topology is challenging. For instance, using RDMA programming features comes at the cost of connection scalability, which does not scale with increasing cluster size. To address that, several works forgo some RDMA features by only focusing on conventional RPC APIs.

Air pollution is a public health emergency in large cities. The availability of commodity sensors and the advent of Internet of Things (IoT) enable the deployment of a city-wide network of 1000's of low-cost real-time air quality monitors to help manage this challenge. This needs to be supported by an IoT cyber-infrastructure for reliable and scalable data acquisition from the edge to the Cloud. The low accuracy of such sensors also motivates the need for data-driven calibration models that can accurately predict the science variables from the raw sensor signals. Here, we offer our experiences with designing and deploying such an IoT software platform and calibration models, and validate it through a pilot field deployment at two mega-cities, Delhi and Mumbai. Our edge data service is able to even-out the differential bandwidths from the sensing devices and to the Cloud repository, and recover from transient failures. Our analytical models reduce the errors of the sensors from a best-case of 63% using the factory baseline to as low as 21%, and substantially advances the state-of-the-art in this domain.

Edge and fog computing have grown popular as IoT deployments become wide-spread. While application composition and scheduling on such resources are being explored, there exists a gap in a distributed data storage service on the edge and fog layer, instead depending solely on the cloud for data persistence. Such a service should reliably store and manage data on fog and edge devices, even in the presence of failures, and offer transparent discovery and access to data for use by edge computing applications. Here, we present ElfStore, a first-of-its-kind edge-local federated store for streams of data blocks. It uses reliable fog devices as a super-peer overlay to monitor the edge resources, offers federated metadata indexing using Bloom filters, locates data within 2-hops, and maintains approximate global statistics about the reliability and storage capacity of edges. Edges host the actual data blocks, and we use a unique differential replication scheme to select edges on which to replicate blocks, to guarantee a minimum reliability and to balance storage utilization. Our experiments on two IoT virtual deployments with 20 and 272 devices show that ElfStore has low overheads, is bound only by the network bandwidth, has scalable performance, and offers tunable resilience.