Resource Disaggregation Research Articles

Low-latency optical switching technology for next-generation edge-cloud computing platforms

We investigated the application of compact and densely integrated silicon photonics (SiP) switches and controlled them using layer 2 labels to achieve low-latency and low-power-consumption data transfer for resource disaggregation and convergence in edge/cloud computing platforms. We utilize virtual LAN identifiers of the standardized Ethernet frame as labels of the packet, which were read by the look-up table (LUT) formed in a field-programmable gate array to control the output destination of the switch. During the label control operation, a delay of 140 ns in the LUT is observed using the software and considering delay in label addition/identification and the off-logic processing delay in transmission/reception, the delay in label addition/identification can be expected to be on the order of microseconds. Three SiP Mach-Zehnder-Interferometer-type switches were monolithically integrated with three inputs and four outputs. The bit error rate (BER) of each switch path was measured, and it was found that the BER was less than 10−8 at all paths.

Towards Full Stack Adaptivity in Permissioned Blockchains

This paper articulates our vision for a learning-based untrustworthy distributed database. We focus on permissioned blockchain systems as an emerging instance of untrustworthy distributed databases and argue that as novel smart contracts, modern hardware, and new cloud platforms arise, future-proof permissioned blockchain systems need to be designed with full-stack adaptivity in mind. At the application level, a future-proof system must adaptively learn the best-performing transaction processing paradigm and quickly adapt to new hardware and unanticipated workload changes on the fly. Likewise, the Byzantine consensus layer must dynamically adjust itself to the workloads, faulty conditions, and network configuration while maintaining compatibility with the transaction processing paradigm. At the infrastructure level, cloud providers must enable cross-layer adaptation, which identifies performance bottlenecks and possible attacks, and determines at runtime the degree of resource disaggregation that best meets application requirements. Within this vision of the future, our paper outlines several research challenges together with some preliminary approaches.

Reinvent Cloud Software Stacks for Resource Disaggregation

Reinvent Cloud Software Stacks for Resource Disaggregation

Make It Real: An End-to-End Implementation of A Physically Disaggregated Data Center

Resource disaggregation is an approach to separate different hardware resources into independent pools in a data center, so that these pools can be easily managed and their resources can be allocated in a tight but unbounded way. The past decade has seen research and practices in realizing the resource-disaggregation idea on regular servers. We advocate for a physically disaggregated data center, where disaggregated resource pools consist of hardware devices, not servers. Physical disaggregation could unlock another level of benefits in resource disaggregation, including further improved cost saving, easier maintenance and scaling, and more customization. This paper presents our efforts in building an end-to-end physically disaggregated data center, including the design and implementation of disaggregated hardware devices, networking systems for connecting these devices, operating systems for orchestrating them, and porting of traditional and cloud-computing applications to this physically disaggregated platform.

Memtrade: Marketplace for Disaggregated Memory Clouds

We present Memtrade, the first practical marketplace for disaggregated memory clouds. Clouds introduce a set of unique challenges for resource disaggregation across different tenants, including resource harvesting, isolation, and matching. Memtrade allows producer virtual machines (VMs) to lease both their unallocated memory and allocated-but-idle application memory to remote consumer VMs for a limited period of time. Memtrade does not require any modifications to host-level system software or support from the cloud provider. It harvests producer memory using an application-aware control loop to form a distributed transient remote memory pool with minimal performance impact; it employs a broker to match producers with consumers while satisfying performance constraints; and it exposes the matched memory to consumers through different abstractions. As a proof of concept, we propose two such memory access interfaces for Memtrade consumers -- a transient KV cache for specified applications and a swap interface that is application-transparent. Our evaluation using real-world cluster traces shows that Memtrade provides significant performance benefit for consumers (improving average read latency up to 2.8X) while preserving confidentiality and integrity, with little impact on producer applications (degrading performance by less than 2.1%).

Exploring the Benefits of Resource Disaggregation for Service Reliability in Data Centers

By overcoming the server box barrier, resource disaggregation in data centers significantly improves resource utilization, which also provides a more cost-efficient approach for resource upgrade and expansion. The advantages of disaggregation have been explored in earlier research to improve the resource efficiency. This paper investigates the potential benefits of disaggregation from the aspect of reliability, which has not been considered before. Resource disaggregation brings a new failure pattern. For example, in a conventional server, the failure of one type of resource leads to the failure of the entire server, so that other types of resources in the same server also become unavailable. After disaggregating, the failure of different types of resources becomes more isolated, so that other resources are still available. In this paper, we model the reliability of a resource allocation request in a server-based or disaggregated DC based on whether the request is allocated with only working resources, or also provisioned with backup. We then consider a resource allocation problem with the objective of maximizing the number of requests accepted with guaranteed reliability. We provide an integer linear programming and a heuristic approach for this. Numerical studies demonstrate that resource disaggregation is possible to improve service reliability.

Memory Pooling With CXL

Compute Express Link (CXL) has recently attracted great attention thanks to its excellent hardware heterogeneity management and resource disaggregation capabilities. Even though there is yet no commercially available product or platform integrating CXL into memory pooling, it is expected to make memory resources practically and efficiently disaggregated much better than ever before. In this article, we propose directly accessible memory disaggregation, DirectCXL that straight connects a host processor complex and remote memory resources over CXL’s memory protocol (CXL.mem). Our empirical evaluation shows that DirectCXL exhibits around 7× better performance than remote direct memory access (RDMA)-based memory pooling for diverse real-world workloads.

DaeMon: Architectural Support for Efficient Data Movement in Fully Disaggregated Systems

Resource disaggregation offers a cost effective solution to resource scaling, utilization, and failure-handling in data centers by physically separating hardware devices in a server. Servers are architected as pools of processor, memory, and storage devices, organized as independent failure-isolated components interconnected by a high-bandwidth network. A critical challenge, however, is the high performance penalty of accessing data from a remote memory module over the network. Addressing this challenge is difficult as disaggregated systems have high runtime variability in network latencies/bandwidth, and page migration can significantly delay critical path cache line accesses in other pages. This paper conducts a characterization analysis on different data movement strategies in fully disaggregated systems, evaluates their performance overheads in a variety of workloads, and introduces DaeMon, the first software-transparent mechanism to significantly alleviate data movement overheads in fully disaggregated systems. First, to enable scalability to multiple hardware components in the system, we enhance each compute and memory unit with specialized engines that transparently handle data migrations. Second, to achieve high performance and provide robustness across various network, architecture and application characteristics, we implement a synergistic approach of bandwidth partitioning, link compression, decoupled data movement of multiple granularities, and adaptive granularity selection in data movements. We evaluate DaeMon in a wide variety of workloads at different network and architecture configurations using a state-of-the-art simulator. DaeMon improves system performance and data access costs by 2.39× and 3.06×, respectively, over the widely-adopted approach of moving data at page granularity.

A Case For Intra-rack Resource Disaggregation in HPC

The expected halt of traditional technology scaling is motivating increased heterogeneity in high-performance computing (HPC) systems with the emergence of numerous specialized accelerators. As heterogeneity increases, so does the risk of underutilizing expensive hardware resources if we preserve today’s rigid node configuration and reservation strategies. This has sparked interest in resource disaggregation to enable finer-grain allocation of hardware resources to applications. However, there is currently no data-driven study of what range of disaggregation is appropriate in HPC. To that end, we perform a detailed analysis of key metrics sampled in NERSC’s Cori, a production HPC system that executes a diverse open-science HPC workload. In addition, we profile a variety of deep-learning applications to represent an emerging workload. We show that for a rack (cabinet) configuration and applications similar to Cori, a central processing unit with intra-rack disaggregation has a 99.5% probability to find all resources it requires inside its rack. In addition, ideal intra-rack resource disaggregation in Cori could reduce memory and NIC resources by 5.36% to 69.01% and still satisfy the worst-case average rack utilization.

Workload-Aware Placement Strategies to Leverage Disaggregated Resources in the Datacenter

Disaggregation of resources is a datacenter strategy that aims to decouple the physical location of resources from the place where they are accessed, as opposed to physically attached devices connected to the Peripheral Component Interconnect Express (PCIe) bus. By attaching and detaching resources through a fast interconnection network, it is possible to increase the flexibility to manage datacenter infrastructures while keeping the performance of the pooled and disaggregated devices. This article introduces workload scheduling and placement policies for environments with disaggregated memories. These policies are driven by accurate prebuilt performance degradation models. We focus on the use of nonvolatile memory to store data and/or to provide memory extensions. Following a software-defined approach, persistent memories are combined to provide higher capacity and/or bandwidth devices, or used by multiple workloads to increase the number of running workloads. Different combinations of workloads and associated soft deadlines are used to evaluate the placement policies using a system simulator. When using the first-fit policy, results show that a disaggregated system can reduce missed deadlines up to 49% when compared to a physically attached system. When our proposed policy with workload awareness is enabled in a disaggregated system, missed deadlines can be reduced up to 100% (no deadlines missed).

A review of spatial resolution and regionalisation in national-scale energy systems optimisation models

National-scale energy systems optimisation models (ESOMs) have been limited in the past by the lack of sub-national data availability leading to aggregated treatment of spatial dynamics. This review paper first determines how a combination of supply and demand data requirements and socio-economic, environmental and political issues, can challenge the results of a low-spatial resolution model. It also demonstrates the incompleteness of single region ESOMs that do not capture sufficient spatial detail. Specifically, 36 multi-regional ESOMs from 22 countries with varying levels of spatiotemporal resolution, sectoral focus and planning horizon are systematically identified and comprehensively analysed. The review reveals that existing temporally explicit ESOMs with a single sector coverage can permit regional disaggregation up to the first-level administrative divisions within a country (such as state and province) while maintaining computationally tractable. Findings from the literature review also show when, how, and to what extent higher spatial resolution impact on the results of energy system analysis. (1) Finer spatial resolution in ESOMs offers significant added value for regions with heterogeneous renewable potential or across regions with higher variability in energy service demands. However, in homogeneous areas, aggregated single-region modelling is more efficient. (2) Spatially resolved models can significantly change the results of the scenarios with very high shares of variable renewable energies. But it is not straightforward to find a direct relationship between the level of geographic disaggregation and penetration of renewable energies. This trade-off should be explored case-by-case. (3) Total system costs can be under- or over-estimated in various levels of spatial resolutions. Disaggregation of renewable resources leads to lower costs, and disaggregation of transmission grids leads to higher costs. • Multi-regional national-scale energy systems optimisation models (ESOMs) reviewed. • Spatiotemporal resolution of multi-regional national ESOMs is compared. • Spatially resolved models offer crucial added value for heterogeneous regions. • The relationship between levels of spatial resolutions and total costs analysed.

MONet: heterogeneous Memory over Optical Network for large-scale data center resource disaggregation

The memory over optical network (MONet) system is a disaggregated data center architecture where serial (HMC)/parallel (DDR4) memory resources can be accessed over optically switched interconnects within and between racks. An FPGA/ASIC-based custom hardware IP (ReMAT) supports heterogeneous memory pools, accommodates optical-to-electrical conversion for remote access, performs the required serial/parallel conversion, and hosts the necessary local memory controller. An optically interconnected HMC-based (serial I/O type) memory card is accessed by a memory controller embedded in the compute card, simplifying the hardware near the memory modules. This substantially reduces overheads on latency, cost, power consumption, and space. We characterize CPU–memory performance by experimentally demonstrating the impact of distance, number of switching hops, transceivers, channel bonding, and bit rate per transceiver on the bit error rate, power consumption, additional latency, sustained remote memory bandwidth/throughput (using industry standard benchmark STREAMS), and cloud workload performance (such as operations per second, average added latency, and retired instructions per second memcached with YCSB cloud workloads). MONet pushes the CPU–memory operational limit from a few centimeters to tens of meters, yet applications can experience as low as 10% performance penalty (at 36 m) compared to a direct-attached equivalent. Using the proposed parallel topology, a system can support up to 100,000 disaggregated cards.

Operating Systems for Resource-adaptive Intelligent Software: Challenges and Opportunities

The past decades witnessed the fast and wide deployment of Internet. The Internet has bred the ubiquitous computing environment that is spanning the cloud, edge, mobile devices, and IoT. Software running over such a ubiquitous computing environment environment is eating the world. A recently emerging trend of Internet-based software systems is “ resource adaptive ,” i.e., software systems should be robust and intelligent enough to the changes of heterogeneous resources, both physical and logical, provided by their running environment. To keep pace of such a trend, we argue that some considerations should be taken into account for the future operating system design and implementation. From the structural perspective, rather than the “monolithic OS” that manages the aggregated resources on the single machine, the OS should be dynamically composed over the distributed resources and flexibly adapt to the resource and environment changes. Meanwhile, the OS should leverage advanced machine/deep learning techniques to derive configurations and policies and automatically learn to tune itself and schedule resources. This article envisions our recent thinking of the new OS abstraction, namely, ServiceOS , for future resource-adaptive intelligent software systems. The idea of ServiceOS is inspired by the delivery model of “ Software-as-a-Service ” that is supported by the Service-Oriented Architecture (SOA). The key principle of ServiceOS is based on resource disaggregation, resource provisioning as a service, and learning-based resource scheduling and allocation. The major goal of this article is not providing an immediately deployable OS. Instead, we aim to summarize the challenges and potentially promising opportunities and try to provide some practical implications for researchers and practitioners.

DRMaestro: orchestrating disaggregated resources on virtualized data-centers

Modern applications demand resources at an unprecedented level. In this sense, data-centers are required to scale efficiently to cope with such demand. Resource disaggregation has the potential to improve resource-efficiency by allowing the deployment of workloads in more flexible ways. Therefore, the industry is shifting towards disaggregated architectures, which enables new ways to structure hardware resources in data centers. However, determining the best performing resource provisioning is a complicated task. The optimality of resource allocation in a disaggregated data center depends on its topology and the workload collocation. This paper presents DRMaestro, a framework to orchestrate disaggregated resources transparently from the applications. DRMaestro uses a novel flow-network model to determine the optimal placement in multiple phases while employing best-efforts on preventing workload performance interference. We first evaluate the impact of disaggregation regarding the additional network requirements under higher network load. The results show that for some applications the impact is minimal, but other ones can suffer up to 80% slowdown in the data transfer part. After that, we evaluate DRMaestro via a real prototype on Kubernetes and a trace-driven simulation. The results show that DRMaestro can reduce the total job makespan with a speedup of up to ≈1.20x and decrease the QoS violation up to ≈2.64x comparing with another orchestrator that does not support resource disaggregation.

Energy Efficient Placement of Workloads in Composable Data Center Networks

This paper studies the energy efficiency of composable data center (DC) infrastructures over network topologies. Using a mixed integer linear programming (MILP) model, we compare the performance of disaggregation at rack-scale and pod-scale over selected electrical, optical and hybrid network topologies relative to a traditional DC. Relative to a pod-scale DC, the results show that physical disaggregation at rack-scale is sufficient for optimal efficiency when the optical network topology is adopted, and resource components are allocated in a suitable manner. The optical network topology also enables optimal energy efficiency in composable DCs. The paper also studies logical disaggregation of traditional DC servers over an optical network topology. Relative to physical disaggregation at rack-scale, logical disaggregation of server resources within each rack enables marginal fall in the total DC power consumption (TDPC) due to improved resource demands placement. Hence, an adaptable composable infrastructure that can support both in memory (access) latency sensitive and insensitive workloads is enabled. We also conduct a study of the adoption of micro-service architecture in both traditional and composable DCs. Our results show that increasing the modularity of workloads improves the energy efficiency in traditional DCs, but disproportionate utilization of DC resources persists. A combination of disaggregation and micro-services achieved up to 23% reduction in the TDPC of the traditional DC by enabling optimal resources utilization and energy efficiencies. Finally, we propose a heuristic for energy efficient placement of workloads in composable DCs which replicates the trends produced by the MILP model formulated in this paper.

Disaggregated Accelerator Management System for Cloud Data Centers

A conventional data center that consists of monolithic-servers is confronted with limitations including lack of operational flexibility, low resource utilization, low maintainability, etc. Resource disaggregation is a promising solution to address the above issues. We propose a concept of disaggregated cloud data center architecture called Flow-in-Cloud (FiC) that enables an existing cluster computer system to expand an accelerator pool through a high-speed network. FlowOS-RM manages the entire pool resources, and deploys a user job on a dynamically constructed slice according to a user request. This slice consists of compute nodes and accelerators where each accelerator is attached to the corresponding compute node. This paper demonstrates the feasibility of FiC in a proof of concept experiment running a distributed deep learning application on the prototype system. The result successfully warrants the applicability of the proposed system.

PINE: Photonic Integrated Networked Energy efficient datacenters (ENLITENED Program) [Invited

We review the motivation, goals, and achievements of the Photonic Integrated Networked Energy efficient datacenter (PINE) project, which is part of the Advanced Research Projects Agency–Energy (ARPA-E) ENergy-efficient Light-wave Integrated Technology Enabling Networks that Enhance Dataprocessing (ENLITENED) program. The PINE program leverages the unique features of photonic technologies to enable alternative mega-datacenters and high-performance computing (HPC) system architectures that deliver more substantial energy efficiency improvements than can be achieved through link energy efficiency alone. In phase 1 of the program, the PINE system architecture demonstrated an average factor of 2.2 × improvement in transactions/joule across a diverse set of HPC and datacenter applications. In phase 2, PINE will demonstrate an aggressive 1.0 pJ/bit total link budget with high-bandwidth-density dense wavelength-division multiplexing (DWDM) links to enable additional 2.5 × or more efficiency gains through deep resource disaggregation.

Cooperative Memory Expansion via OS Kernel Support for Networked Computing Systems

The growing popularity of in-memory computing for bigdata analytics often causes performance bottlenecks to memory subsystem resided in operating systems (OS). This article purposes cooperative memory expansion (COMEX), an OS kernel extension. COMEX establishes a stable pool of memory collectively across nodes in a cluster and enhances OS's memory subsystem for memory aggregation from connected machines by allowing process's page table to track remote memory page frames without programmer effort or modifications to application codes. COMEX employs Remote Direct Memory Access (RDMA) for low-latency data transfer with destination kernel bypassed and does not rely on an old design of the I/O block subsystem usually adopted by all known remote paging. COMEX fits soundly in the emerging system design approach of resource disaggregation which breaks hard walls between server-centric machines into a new design paradigm of separated resource pools. The new architecture facilitates both system scaling-up and scaling-out, also eliminates imbalance resources existing in datacenters. We have implemented COMEX based on Linux kernel 3.10.87 and deployed on our 32 networked servers. Performance evaluation results under ten applications from two benchmark suites reveal the speedup of up to 170 times when application execution footprints are 10 times larger than available system memory.

Optimized placement of virtualized resources for 5G services exploiting live migration

This paper focuses on a Centralized Radio Access Network solution adopting the concept of resource disaggregation. In this context, it proposes a heuristic suitable to optimally assign Base Band Unit processing functions in softwarized Radio Access Networks to different servers, taking into consideration their processing requirements with the aim to minimize the overall energy consumption. It also proposes the adoption of live migration of virtualized resources, in order to dynamically reallocate these functions to different servers that better match the continuously changing characteristics of 5G services, for increased energy efficiency purposes. The benefits associated with live migration are quantified through a series of experiments. Our results show a reduction of the number of switched-on servers through live migration that leads to a notable improvement in terms of resource and energy efficiency.

Understanding the effect of data center resource disaggregation on production DBMSs

Resource disaggregation is a new architecture for data centers in which resources like memory and storage are decoupled from the CPU, managed independently, and connected through a high-speed network. Recent work has shown that although disaggregated data centers (DDCs) provide operational benefits, applications running on DDCs experience degraded performance due to extra network latency between the CPU and their working sets in main memory. DBMSs are an interesting case study for DDCs for two main reasons: (1) DBMSs normally process data-intensive workloads and require data movement between different resource components; and (2) disaggregation drastically changes the assumption that DBMSs can rely on their own internal resource management. We take the first step to thoroughly evaluate the query execution performance of production DBMSs in disaggregated data centers. We evaluate two popular open-source DBMSs (MonetDB and PostgreSQL) and test their performance with the TPC-H benchmark in a recently released operating system for resource disaggregation. We evaluate these DBMSs with various configurations and compare their performance with that of single-machine Linux with the same hardware resources. Our results confirm that significant performance degradation does occur, but, perhaps surprisingly, we also find settings in which the degradation is minor or where DDCs actually improve performance.