Heterogeneous memory combining DRAM and CXL exhibits variable performance, yet existing metrics correlate weakly with actual slowdown. We present CAMP, a principled framework for predicting CXL-induced slowdown. Our key insight is that a DRAM run (plus a CXL run for bandwidth-bound workloads) exposes the causal microarchitectural pressure points where CXL latency translates into additional processor stall cycles. CAMP captures these signals using 12 performance counters to analytically decompose slowdown into three orthogonal components: demand reads, cache/prefetching, and stores. CAMP also introduces a closed-form model for software-based weighted interleaving that predicts performance across DRAM--CXL ratios. Across 265 workloads on NUMA and three CXL devices, CAMP achieves 91--97% prediction accuracy within 10% absolute error. We demonstrate that these models enable practical system policies, including ''Best-shot'' interleaving and colocated workload placement, improving performance by up to 21% and 23% over existing tiering and colocation approaches.

This paper explores learnings from first-generation Compute Express Link (CXL) memory expansion to accelerate CXL’s journey to broad, robust use. While broad adoption will be a long journey similar to that of RDMA, we argue that the first step—CXL.mem expansion—is viable on today’s hardware. Through an end-to-end analysis, we revisit common showstoppers: we decompose memory access latency and show that CPU and DRAM internals, rather than the CXL protocol, dominate latency and variability, and we demonstrate how system slack absorbs link error rates above nominal specifications. Along the way, we distill practical guidance on device validation, monitoring, failure modes, security, and multi-tenant interference, and we outline a pragmatic adoption pathway: solidify robust expansion first, prototype micro-pooling next, and move to selective sharing as the ecosystem matures.

Compute Express Link (CXL) has emerged as a pivotal interconnect for memory expansion. Despite its potential, the performance implications of CXL across devices, latency regimes, processors, and workloads remain underexplored. We present Melody, a framework for systematic characterization and analysis of CXL memory performance. Melody builds on an extensive evaluation spanning 265 workloads, 4 real CXL devices, 7 latency levels, and 5 CPU platforms. Melody yields many insights: workload sensitivity to sub-μs CXL latencies (140-410ns), the first disclosure of CXL tail latencies, CPU tolerance to CXL latencies, a novel approach (SPA) for pinpointing CXL bottlenecks, and CPU prefetcher inefficiencies under CXL.

Cloud platforms remain underutilized despite multiple proposals to improve their utilization (e.g., disaggregation, harvesting, and oversubscription). Our characterization of the resource utilization of virtual machines (VMs) in Azure reveals that, while CPU is the main underutilized resource, we need to provide a solution to manage all resources holistically. We also observe that many VMs exhibit complementary temporal patterns, which can be leveraged to improve the oversubscription of underutilized resources.

Pooling PCIe devices across multiple hosts offers a promising solution to mitigate stranded I/O resources, enhance device utilization, address device failures, and reduce total cost of ownership. The only viable option today are PCIe switches, which decouple PCIe devices from hosts by connecting them through a hardware switch. However, the high cost and limited flexibility of PCIe switches hinder their widespread adoption beyond specialized datacenter use cases.

PCIe devices, such as NICs and SSDs, are frequently underutilized in cloud platforms. PCIe device pools, in which multiple hosts can share a set of PCIe devices, could increase PCIe device utilization and reduce their total cost of ownership. The main way to achieve PCIe device pools today is via PCIe switches, but they are expensive and inflexible. We design Oasis,1 a system that pools PCIe devices in software over CXL memory pools. CXL memory pools are already being deployed to boost datacenter memory utilization and reduce costs. Once CXL pools are in place, they can serve as an efficient data path between hosts and PCIe devices. Oasis provides a control plane and datapath over CXL pools, mapping and routing PCIe device traffic across host boundaries. PCIe devices with different functionalities can be supported by adding an Oasis engine for each device class. We implement an Oasis network engine to demonstrate NIC pooling. Our evaluation shows that Oasis improves the NIC utilization by 2× and handles NIC failover with only a 38 ms interruption.

Datacenters need to reduce embodied carbon emissions, particularly for flash, which accounts for 40% of embodied carbon in servers. However, decreasing flash’s embodied emissions is challenging due to flash’s limited write endurance, which more than halves with each generation of denser flash. Reducing embodied emissions requires extending flash lifetime, stressing its limited write endurance even further. The legacy Logical Block-Addressable Device (LBAD) interface exacerbates the problem by forcing devices to perform garbage collection, leading to even more writes. Flash-based caches in particular write frequently, limiting the lifetimes and densities of the devices they use. These flash caches illustrate the need to break away from LBAD and switch to the new Write-Read-Erase iNterfaces (WREN) now coming to market. WREN affords applications control over data placement and garbage collection. We present Fairy Wren , 1 a flash cache designed for WREN. Fairy Wren reduces writes by co-designing caching policies and flash garbage collection. Fairy Wren provides a 12.5× write reduction over state-of-the-art LBAD caches. This decrease in writes allows flash devices to last longer, decreasing flash cost by 35% and flash carbon emissions by 33%.

To mitigate climate change, we must reduce carbon emissions from hyperscale cloud computing. We find that cloud compute servers cause the majority of emissions in a general-purpose cloud. Thus, we motivate designing carbon-efficient compute server SKUs, or GreenSKUs, using recently-available low-carbon server components. To this end, we design and build three GreenSKUs using low-carbon components, such as energy-efficient CPUs, reused old DRAM via CXL, and reused old SSDs.We detail several challenges that limit GreenSKUs, carbon savings at scale and may prevent their adoption by cloud providers. To address these challenges, we develop a novel methodology and associated framework, GSF (GreenSKU Framework), that enables a cloud provider to systematically evaluate a GreenSKU’s carbon savings at scale. We implement GSF within Microsoft Azure’s production constraints to evaluate our three GreenSKUs’ carbon savings. Using GSF, we show that our most carbon-efficient GreenSKU reduces emissions per core by $28 \%$ compared to currently-deployed cloud servers. When designing GreenSKUs to meet applications’ performance requirements, we reduce emissions by $15 \%$. When incorporating overall data center overheads, our GreenSKU reduces Azure’s net cloud emissions by $8 \%$.

The growing demands for computational power in cloud computing have led to a significant increase in the deployment of high-performance servers. The growing power consumption of servers and the heat they produce is on track to outpace the capacity of conventional air cooling systems, necessitating more efficient cooling solutions such as liquid immersion cooling. The superior heat exchange capabilities of immersion cooling both eliminates the need for bulky heat sinks, fans, and air flow channels while also unlocking the potential go beyond conventional 2D blade servers to three-dimensional designs. In this work, we present a computational framework to explore designs of servers in three-dimensional space, specifically targeting the maximization of server density within immersion cooling tanks. Our tool is designed to handle a variety of physical and electrical server design constraints. We demonstrate our optimized designs can reduce server volume by 25--52% compared to traditional flat server designs. This increased density reduces land usage as well as the amount of liquid used for immersion, with significant reduction in the carbon emissions embodied in datacenter buildings. We further create physical prototypes to simulate dense server designs and perform real-world experiments in an immersion cooling tank demonstrating they operate at safe temperatures. This approach marks a critical step forward in sustainable and efficient datacenter management.

Public cloud providers seek to meet stringent performance requirements and low hardware cost. A key driver of performance and cost is main memory. Memory pooling promises to improve DRAM utilization and thereby reduce costs. However, pooling is challenging under cloud performance requirements. This paper proposes Pond, the first memory pooling system that both meets cloud performance goals and significantly reduces DRAM cost. Pond builds on the Compute Express Link (CXL) standard for load/store access to pool memory and two key insights. First, our analysis of cloud production traces shows that pooling across 8-16 sockets is enough to achieve most of the benefits. This enables a small-pool design with low access latency. Second, it is possible to create machine learning models that can accurately predict how much local and pool memory to allocate to a virtual machine (VM) to resemble same-NUMA-node memory performance. Our evaluation with 158 workloads shows that Pond reduces DRAM costs by 7% with performance within 1-5% of same-NUMA-node VM allocations.

Function-as-a-Service (FaaS) serverless computing enables a simple programming model with almost unbounded elasticity. Unfortunately, current FaaS platforms achieve this flexibility at the cost of lower performance for data-intensive applications compared to a serverful deployment. The ability to have computation close to data is a key missing feature. We introduce Palette load balancing, which offers FaaS applications a simple mechanism to express locality to the platform, through hints we term "colors". Palette maintains the serverless nature of the service - users are still not allocating resources - while allowing the platform to place successive invocations related to each other on the same executing node. We compare a prototype of the Palette load balancer to a state-of-the-art locality-oblivious load balancer on representative examples of three applications. For a serverless web application with a local cache, Palette improves the hit ratio by 6x. For a serverless version of Dask, Palette improves run times by 46% and 40% on Task Bench and TPC-H, respectively. On a serverless version of NumS, Palette improves run times by 37%. These improvements largely bridge the gap to serverful implementation of the same systems.

Cloud platforms run many software agents on each server node. These agents manage all aspects of node operation, and in some cases frequently collect data and make decisions. Unfortunately, their behavior is typically based on pre-defined static heuristics or offline analysis; they do not leverage on-node machine learning (ML). In this paper, we first characterize the spectrum of node agents in Azure, and identify the classes of agents that are most likely to benefit from on-node ML. We then propose SOL, an extensible framework for designing ML-based agents that are safe and robust to the range of failure conditions that occur in production. SOL provides a simple API to agent developers and manages the scheduling and running of the agent-specific functions they write. We illustrate the use of SOL by implementing three ML-based agents that manage CPU cores, node power, and memory placement. Our experiments show that (1) ML substantially improves our agents, and (2) SOL ensures that agents operate safely under a variety of failure conditions. We conclude that ML-based agents show significant potential and that SOL can help build them.

Many social-media and IoT services have very large working sets consisting of billions of tiny (≈100 B) objects. Large, flash-based caches are important to serving these working sets at acceptable monetary cost. However, caching tiny objects on flash is challenging for two reasons: (i) SSDs can read/write data only in multi-KB “pages” that are much larger than a single object, stressing the limited number of times flash can be written; and (ii) very few bits per cached object can be kept in DRAM without losing flash’s cost advantage. Unfortunately, existing flash-cache designs fall short of addressing these challenges: write-optimized designs require too much DRAM, and DRAM-optimized designs require too many flash writes. We present Kangaroo , a new flash-cache design that optimizes both DRAM usage and flash writes to maximize cache performance while minimizing cost. Kangaroo combines a large, set-associative cache with a small, log-structured cache. The set-associative cache requires minimal DRAM, while the log-structured cache minimizes Kangaroo’s flash writes. Experiments using traces from Meta and Twitter show that Kangaroo achieves DRAM usage close to the best prior DRAM-optimized design, flash writes close to the best prior write-optimized design, and miss ratios better than both. Kangaroo’s design is Pareto-optimal across a range of allowed write rates, DRAM sizes, and flash sizes, reducing misses by 29% over the state of the art. These results are corroborated by analytical models presented herein and with a test deployment of Kangaroo in a production flash cache at Meta.

Continuously monitoring a wide variety of performance and fault metrics has become a crucial part of operating large-scale datacenter networks. In this work, we ask whether we can reduce the costs to monitor - in terms of collection, storage and analysis - by judiciously controlling how much and which measurements we collect. By positing that we can treat almost all measured signals as sampled time-series, we show that we can use signal processing techniques such as the Nyquist-Shannon theorem to avoid wasteful data collection. We show that large savings appear possible by analyzing tens of popular measurement systems from a production datacenter network. We also discuss some challenges that must be solved when applying these techniques in practice.

Kernel-bypass networking, which allows applications to circumvent the kernel and interface directly with NIC hardware, is one of the main tools for improving application network performance. However, allowing applications to circumvent the kernel makes it impossible to use tools (e.g., tcpdump) or impose policies (e.g., QoS and filters) that need to interpose on traffic sent by different applications running on a host. This makes maintainability and manageability a challenge for kernel-bypass applications. In response, we propose Kernel On-Path Interposition (KOPI), in which traditional kernel data-plane functionality is retained but implemented in a fully programmable SmartNIC. We hypothesize that KOPI can support the same tools and policies as the kernel stack while retaining the performance benefits of kernel bypass.

Research on flash devices almost exclusively focuses on conventional SSDs, which expose a block interface. Industry, however, has standardized and is adopting Zoned Namespaces (ZNS) SSDs, which offer a new storage interface that dominates conventional SSDs. Continued research on conventional SSDs is thus a missed opportunity to unlock a step-change improvement in system performance by building on ZNS SSDs. We argue for an immediate and complete shift in research to ZNS SSDs and discuss research directions.

Many social-media and IoT services have very large working sets consisting of billions of tiny (≈100 B) objects. Large, flash-based caches are important to serving these working sets at acceptable monetary cost. However, caching tiny objects on flash is challenging for two reasons: (i) SSDs can read/write data only in multi-KB "pages" that are much larger than a single object, stressing the limited number of times flash can be written; and (ii) very few bits per cached object can be kept in DRAM without losing flash's cost advantage. Unfortunately, existing flash-cache designs fall short of addressing these challenges: write-optimized designs require too much DRAM, and DRAM-optimized designs require too many flash writes.

Redy is a cloud service that provides high performance caches using RDMA-accessible remote memory. An application can customize the performance of each cache with a service level objective (SLO) for latency and throughput. By using remote memory, it can leverage stranded memory and spot VM instances to reduce the cost of its caches and improve data center resource utilization. Redy automatically customizes the resource configuration for the given SLO, handles the dynamics of remote memory regions, and recovers from failures. The experimental evaluation shows that Redy can deliver its promised performance and robustness under remote memory dynamics in the cloud. We augment a production key-value store, FASTER, with a Redy cache. When the working set exceeds local memory, using Redy is significantly faster than spilling to SSDs.

Caches are at the heart of latency-sensitive systems. In this paper, we identify a growing challenge for the design of latency-minimizing caches called delayed hits. Delayed hits occur at high throughput, when multiple requests to the same object queue up before an outstanding cache miss is resolved. This effect increases latencies beyond the predictions of traditional caching models and simulations; in fact, caching algorithms are designed as if delayed hits simply didn't exist. We show that traditional caching strategies -- even so called 'optimal' algorithms -- can fail to minimize latency in the presence of delayed hits. We design a new, latency-optimal offline caching algorithm called belatedly which reduces average latencies by up to 45% compared to the traditional, hit-rate optimal Belady's algorithm. Using belatedly as our guide, we show that incorporating an object's 'aggregate delay' into online caching heuristics can improve latencies for practical caching systems by up to 40%. We implement a prototype, Minimum-AggregateDelay (mad), within a CDN caching node. Using a CDN production trace and backends deployed in different geographic locations, we show that mad can reduce latencies by 12-18% depending on the backend RTTs.

Recent advances in the field of reinforcement learning promise a general approach to optimize networking systems. This paper argues against the recent trend for generalization by introducing a case study where domain-specific modeling enables the application of lightweight and robust learning techniques.

Meeting tail latency Service Level Objectives (SLOs) in shared cloud networks is both important and challenging. One primary challenge is determining limits on the multi-tenancy such that SLOs are met. Doing so involves estimating latency, which is difficult, especially when tenants exhibit bursty behavior as is common in production environments. Nevertheless, recent papers in the past two years (Silo, QJump, and PriorityMeister) show techniques for calculating latency based on a branch of mathematical modeling called Deterministic Network Calculus (DNC). The DNC theory is designed for adversarial worst-case conditions, which is sometimes necessary, but is often overly conservative. Typical tenants do not require strict worst-case guarantees, but are only looking for SLOs at lower percentiles (e.g., 99th, 99.9th).

Frequency jamming is known as an efficient attack tool to disrupt wireless communication. This efficiency can also be exploited for the benefit of a network—an idea often referred to as friendly jamming. A prominent application case is the blocking of unauthenticated or malicious communication, such as injection attacks. In this paper, we propose access points as a natural place to implement friendly jamming functionality. We analyze this proposal using simulations, introduce an implementation on customer-grade access points, and report measurement results from the first real-world study of friendly jamming in an IEEE 802.11 campus network. We discover a fundamental tradeoff between the effectiveness of friendly jamming and the orthogonal aspect of having minimal side-effects to the campus network’s traffic. In particular, we observed what we call the power amplification phenomenon. This effect aggravates the known hidden station problem when the number of jammers increases. We also find evidence that the collaboration between jammers can enable friendly jamming, which is both effective and minimally invasive.

This poster presents the idea of exploiting mobility to improve the security in location and track verification. Unlike traditional approaches which require tight time synchronization or two-way communication, mobility can be used to derive lightweight verification schemes. By ensuring independent movement of the verifiers, our scheme can provide security guarantees even if the verifiers' positions are known to the attacker. We also give an outlook on more general opportunities for mobility-aided security.

TTL caching models have recently regained significant research interest, largely due to their ability to fit popular caching policies such as LRU. In this extended abstract we briefly describe our recent work on two exact methods to analyze TTL cache networks. The first method generalizes existing results for line networks under renewal requests to the broad class of caching policies whereby evictions are driven by stopping times. The obtained results are further generalized, using the second method, to feedforward networks with Markov arrival processes (MAP) requests. MAPs are particularly suitable for non-line networks because they are closed not only under superposition and splitting, as known, but also under input-output caching operations as proven herein for phase-type TTL distributions. The crucial benefit of the two closure properties is that they jointly enable the first exact analysis of feedforward networks of TTL caches in great generality.

Adversarial Queueing Theory (AQT) has shown that seemingly innocent traffic injection rates might lead to unbounded queues in packet-switched networks - depending on scheduling strategies as well as topological characteristics. Little attention has been given to quantifying these effects in realistic network configurations. In particular, the existing AQT literature makes two unrealistic assumptions: infinite buffers and perfect synchrony. Because finite buffers inherently limit queue sizes, adversarial effects ultimately lead to packet loss which we address in this work. In addition, we study the effect of imperfect network synchronization under the packet loss metric. Our results, using analysis and simulation, indicate that classical AQT examples appear harmless under realistic assumptions but for a novel class of adversaries considerably higher loss can be observed. We introduce this class by giving examples of two new AQT concepts to construct loss-efficient network adversaries. Our analysis proves the robustness of these new adversaries against randomized de-synchronization effects in terms of variable link delays and nodal processing.

Frequency jamming is the fiercest attack tool to disrupt wireless communication and its malicious aspects have received much attention in the literature. Yet, several recent works propose to turn the table and employ so-called friendly jamming for the benefit of a wireless network. For example, recently proposed friendly jamming applications include hiding communication channels, injection attack defense, and access control. This work investigates the practical viability of friendly jamming by applying it in a real-world network. To that end, we implemented a reactive and frame-selective jammer on a consumer grade IEEE 802.11 access point. Equipped with this, we conducted a three weeks real-world study on the jammer's performance and side-effects on legitimate traffic (the cost of jamming) in a university office environment. Our results provide detailed insights on crucial factors governing the trade-off between the effectiveness of friendly jamming (we evaluated up to 13 jammers) and its cost. In particular, we observed -- what we call the power amplification phenomenon -- an effect that aggravates the known hidden station problem when the number of jammers increases. However, we also find evidence that this effect can be alleviated by collaboration between jammers, which again enables effective and minimally invasive friendly jamming.