AI clusters today are one of the major uses of High Bandwidth Memory (HBM). However, HBM is suboptimal for AI workloads for several reasons. Analysis shows HBM is overprovisioned on write performance, but underprovisioned on density and read bandwidth, and also has significant energy per bit overheads. It is also expensive, with lower yield than DRAM due to manufacturing complexity. We propose a new memory class: Managed-Retention Memory (MRM), which is more optimized to store key data structures for AI inference workloads. We believe that MRM may finally provide a path to viability for technologies that were originally proposed to support Storage Class Memory (SCM). These technologies traditionally offered long-term persistence (10+ years) but provided poor IO performance and/or endurance. MRM makes different trade-offs, and by understanding the workload IO patterns, MRM foregoes long-term data retention and write performance for better potential performance on the metrics important for these workloads.

Sustainable and cost-effective long-term storage remains an unsolved problem. The most widely used storage technologies today are magnetic (hard disk drives and tape). They use media that degrades over time and has a limited lifetime, which leads to inefficient, wasteful, and costly solutions for long-lived data. This article presents Silica: the first cloud storage system for archival data underpinned by quartz glass, an extremely resilient media that allows data to be left in situ indefinitely. The hardware and software of Silica have been co-designed and co-optimized from the media up to the service level with sustainability as a primary objective. The design follows a cloud-first, data-driven methodology underpinned by principles derived from analyzing the archival workload of a large public cloud service. Silica can support a wide range of archival storage workloads and ushers in a new era of sustainable, cost-effective storage.

Automated storage and retrieval systems (ASRS) are a key component of the modern storage industry, and are used in a wide range of applications, carrying anything from lightweight tape cartridges to entire pallets of goods. Many of these systems are under pressure to maximise the use of space by growing in height and density, but this can create challenges for the the robots that service them. In this context, we present RASCAL, a novel ASRS robot for small payload items in structured environments, with a focus on system-level scalability and redundancy. We describe the design objectives of RASCAL and how they address some of the limitations of existing robotic systems in this area, such as scalability and redundancy. We then demonstrate the viability of our design with a proof-of-concept implementation of a data centre storage media robot, and show through a series of experiments that its design, speed, accuracy, and energy efficiency are appropriate for this application.

Sustainable and cost-effective long-term storage remains an unsolved problem. The most widely used storage technologies today are magnetic (hard disk drives and tape). They use media that degrades over time and has a limited lifetime, which leads to inefficient, wasteful, and costly solutions for long-lived data. This paper presents Silica: the first cloud storage system for archival data underpinned by quartz glass, an extremely resilient media that allows data to be left in situ indefinitely. The hardware and software of Silica have been co-designed and co-optimized from the media up to the service level with sustainability as a primary objective. The design follows a cloud-first, data-driven methodology underpinned by principles derived from analyzing the archival workload of a large public cloud service. Silica can support a wide range of archival storage workloads and ushers in a new era of sustainable, cost-effective storage.

In data centers, the IO path to storage is long and complex. It comprises many layers or "stages" with opaque interfaces between them. This makes it hard to enforce end-to-end policies that dictate a storage IO flow's performance (e.g., guarantee a tenant's IO bandwidth) and routing (e.g., route an untrusted VM's traffic through a sanitization middlebox). These policies require IO differentiation along the flow path and global visibility at the control plane. We design IOFlow, an architecture that uses a logically centralized control plane to enable high-level flow policies. IOFlow adds a queuing abstraction at data-plane stages and exposes this to the controller. The controller can then translate policies into queuing rules at individual stages. It can also choose among multiple stages for policy enforcement.

Bugs in kernel extensions remain one of the main causes of poor operating system reliability despite proposed techniques that isolate extensions in separate protection domains to contain faults. We believe that previous fault isolation techniques are not widely used because they cannot isolate existing kernel extensions with low overhead on standard hardware. This is a hard problem because these extensions communicate with the kernel using a complex interface and they communicate frequently. We present BGI (Byte-Granularity Isolation), a new software fault isolation technique that addresses this problem. BGI uses efficient byte-granularity memory protection to isolate kernel extensions in separate protection domains that share the same address space. BGI ensures type safety for kernel objects and it can detect common types of errors inside domains. Our results show that BGI is practical: it can isolate Windows drivers without requiring changes to the source code and it introduces a CPU overhead between 0 and 16%. BGI can also find bugs during driver testing. We found 28 new bugs in widely used Windows drivers.

Existing methods for local area service discovery either don't scale or rely on a trustworthy directory server; in some environments these restrictions are unacceptable or impractical. This paper describes "Repeat-BAND", a generic method for service discovery that scales automatically without depending on any central component. The automatic scaling makes it fast on small networks and automatically load controlled on large networks, irrespective of the number of simultaneous discoveries taking place. It is generic in the sense that the automatic scaling technique can be applied to improve any particular service discovery system, and it is applicable across a large variety of types of network because we show how all the tuning parameters are derived. We present results showing controlled load discovery with scalability up to very large networks. We also show that the algorithms are simple and easy to implement; an important practical requirement since the method is used in Windows Vista and licensed by many hardware vendors. In addition, we consider the industrial requirement as to the certification of independent implementations, an aspect normally ignored in the academic literature.

This work addresses the problem of layer 2 topology discovery. Current techniques concentrate on using SNMP to query information from Ethernet switches. In contrast, we present a technique that infers the Ethernet (layer 2) topology without assistance from the network elements by injecting suitable probe packets from the end-systems and observing where they are delivered. We describe the algorithm, formally characterize its correctness and completeness, and present our implementation and experimental results. Performance results show that although originally aimed at the home and small office the techniques scale to much larger networks.

In the management of a modern LAN or campus network, two issues are of key importance, namely network performance and capacity planning. In this paper we report on results from an experimental programme which aims to quantify the performance that can be achieved with a real distributed application running over a range of different network technologies, including Ethernet, ATM, FDDI, and a 100 Mb/s packet-switched LAN. In particular we analyse the contribution of each of the various application and network components to the overall performance experienced by applications.

A vertically structured operating system is one in which neither the "kernel" nor "servers" perform work on behalf of applications-the former because it exists only to multiplex the CPU, and the latter in order to avoid Quality of Service interference between the applications. Instead, wherever possible, the applications perform all of their own processing. Such a vertical structure provides many advantages for applications but leads to some interesting problems and opportunities for protocol stack implementation. This paper describes the techniques we used in our protocol implementation and the benefits that the vertical structure provided.

In the management of a modern LAN or campus network, two issues are of key importance, namely network performance and capacity planning. In this paper we report on results from an experimental programme which aims to quantify the performance that can be achieved with a real distributed application running over a range of different network technologies, including Ethernet, ATM, FDDI, and a 100 Mb/s packet-switched LAN. In particular we analyse the contribution of each of the various application and network components to the overall performance experienced by applications. This information will facilitate planning and management of new networks, and permit the identification of "hidden costs" which prevent currently deployed applications from fully exploiting high bandwidth networks.

Support for multimedia applications by general purpose computing platforms has been the subject of considerable research. Much of this work is based on an evolutionary strategy in which small changes to existing systems are made. The approach adopted is to start ab initio with no backward compatibility constraints. This leads to a novel structure for an operating system. The structure aims to decouple applications from one another and to provide multiplexing of all resources, not just the CPU, at a low level. The motivation for this structure, a design based on the structure, and its implementation on a number of hardware platforms is described.