Carnegie Mellon University

Displaying 1-40 of 536 results

  • White Papers // Feb 2014

    SpringFS: Bridging Agility and Performance in Elastic Distributed Storage

    The elastic storage systems can be expanded or contracted to meet current demand, allowing servers to be turned off or used for other tasks. However, the usefulness of an elastic distributed storage system is limited by its agility: how quickly it can increase or decrease its number of servers. Due...

    Provided By Carnegie Mellon University

  • White Papers // Jan 2014

    Toward Strong, Usable Access Control for Shared Distributed Data

    As non-expert users produce increasing amounts of personal digital data, usable access control becomes critical. Current approaches often fail, because they insufficiently protect data or confuse users about policy specification. This paper presents penumbra, a distributed file system with access control designed to match users' mental models while providing principled...

    Provided By Carnegie Mellon University

  • White Papers // Jan 2014

    ARP Attacks in Wireless Ad Hoc Networks

    Previous research to secure ad hoc network protocols concentrates on key distribution and on securing the routing protocol. Unfortunately, securing the routing layer is not sufficient to secure the ad hoc network, as lower layer and upper layer communication protocols remain vulnerable to attack. In this paper, the authors illustrate...

    Provided By Carnegie Mellon University

  • White Papers // Jan 2014

    Matching Database Access Patterns to Storage Characteristics

    Database storage managers employ sophisticated algorithms attempting to exploit the performance available inside today's storage systems. However, because the communication between the Storage Manager (SM) and a storage device is limited by a high-level protocol, both the SM and the device make decisions largely in isolation and do not realize...

    Provided By Carnegie Mellon University

  • White Papers // Dec 2013

    Tetrisched: Space-Time Scheduling for Heterogeneous Datacenters

    Tetrisched is a new scheduler that explicitly considers both job-specific preferences and estimated job runtimes in its allocation of resources. Combined, this information allows tetrisched to provide higher overall value to complex application mixes consolidated on heterogeneous collections of machines. Job-specific preferences, provided by tenants in the form of composable...

    Provided By Carnegie Mellon University

  • White Papers // Nov 2013

    More Effective Distributed ML via a Stale Synchronous Parallel Parameter Server

    The authors propose a parameter server system for distributed ML, which follows a Stale Synchronous Parallel (SSP) model of computation that maximizes the time computational workers spend doing useful work on ML algorithms, while still providing correctness guarantees. The parameter server provides an easy-to-use shared interface for read/write access to...

    Provided By Carnegie Mellon University

  • Webcasts // Nov 2013

    Illicit Cyber Activity Involving Fraud

    In this webcast, the presenter going to explain about illicit cyber activity involving fraud from the virtual event managing the insider threat. the presenters aim is to assist organizations in improving their security posture and incident response capability by researching technical threat areas, developing and conducting information security assessments, and...

    Provided By Carnegie Mellon University

  • Webcasts // Nov 2013

    Engineering Realistic Synthetic Insider Threat (Cyber-Social) Test Data

    Access to data remains a significant impediment to advancing the science of insider threat research. In this webcast, the presenter going to explain about engineering realistic synthetic about the insider threat test data from the virtual event who manages the insider threat.

    Provided By Carnegie Mellon University

  • Webcasts // Nov 2013

    Emerging Trends

    An emerging trend to pay attention to is the malicious outsider taking advantage of an inadvertent insider. In this webcast, the presenter going to explain about emerging trends in insider threat research from the virtual event managing the insider threat.

    Provided By Carnegie Mellon University

  • Webcasts // Nov 2013

    Best Practices and Controls for Mitigating Insider Threats

    Threats can stem from simple activities like uploading, emailing, and posting sensitive data, etc. In this webcast, the presenter will explain about best practices and controls for mitigating insider threats from the virtual event of how to manage the insider threat.

    Provided By Carnegie Mellon University

  • Webcasts // Nov 2013

    Panel Discussion: Managing the Insider Threat: What Every Organization Should Know

    In this webcast, the presenter going to discuss about managing the insider threat and how to ensure the protection of and appropriate access to intellectual property and other critical assets, systems, and data.

    Provided By Carnegie Mellon University

  • Podcasts // Nov 2013

    Overview of the Threat Posed by Insiders to Critical Assets

    Insiders pose a substantial threat by virtue of their knowledge of and access to the users systems and/or their databases. In this webcast, the presenter will give an overview of the threat posed by insiders to critical assets.

    Provided By Carnegie Mellon University

  • White Papers // Oct 2013

    GOTCHA Password Hackers!

    The authors introduce GOTCHAs (Generating panOptic Turing tests to tell Computers and Humans Apart) as a way of preventing automated offline dictionary attacks against user selected passwords. A GOTCHA is a randomized puzzle generation protocol, which involves interaction between a computer and a human. Informally, a GOTCHA should satisfy two...

    Provided By Carnegie Mellon University

  • White Papers // Oct 2013

    Challenges in Security and Privacy for Mobile Edge-Clouds

    Mobile devices such as Smartphone and tablets are ubiquitous today, and many of them possess significant computation power, powerful sensors such as high-resolution cameras and GPS sensors, and a wealth of sensor data such as photos, videos, and location information. Collections of mobile devices in close geographical proximity present both...

    Provided By Carnegie Mellon University

  • White Papers // Sep 2013

    LightTx: A Lightweight Transactional Design in Flash-based SSDs to Support Flexible Transactions

    Flash memory has accelerated the architectural evolution of storage systems with its unique characteristics compared to magnetic disks. The no-overwrite property of flash memory has been leveraged to efficiently support transactions, a commonly used mechanism in systems to provide consistency. However, existing transaction designs embedded in flash-based Solid State Drives...

    Provided By Carnegie Mellon University

  • White Papers // Sep 2013

    Program Interference in MLC NAND Flash Memory: Characterization, Modeling, and Mitigation

    As NAND flash memory continues to scale down to smaller process technology nodes, its reliability and endurance are degrading. One important source of reduced reliability is the phenomenon of program interference: when a flash cell is programmed to a value, the programming operation affects the threshold voltage of not only...

    Provided By Carnegie Mellon University

  • White Papers // Sep 2013

    HAT: Heterogeneous Adaptive Throttling for On-Chip Networks

    The Network-on-Chip (NoC) is a primary shared resource in a Chip Multi-Processor (CMP) system. As core counts continue to increase and applications become increasingly data-intensive, the network load will also increase, leading to more congestion in the network. This network congestion can degrade system performance if the network load is...

    Provided By Carnegie Mellon University

  • White Papers // Sep 2013

    Secure Two-Party Computation with Reusable Bit-Commitments, via A Cut-and-Choose with Forge-and-Lose Technique

    A Secure Two-Party Computation (S2PC) protocol allows two parties to compute over their combined private inputs, as if inter-mediated by a trusted third party. In the active model, security is maintained even if one party is malicious, deviating from the protocol specification. For example, an honest party retains privacy of...

    Provided By Carnegie Mellon University

  • White Papers // Aug 2013

    Design of a Write-Optimized Data Store

    The WriteBuffer (WB) tree is a new write-optimized data structure that can be used to implement per-node storage in unordered key-value stores. The WB tree provides faster writes than the Log-Structured Merge (LSM) tree that is used in many current high-performance key-value stores. It achieves this by replacing compactions in...

    Provided By Carnegie Mellon University

  • Webcasts // Aug 2013

    Observations of Successful Cyber Security Operations

    In this Webcast, the presenter going explain about the cyber security The presenter will also discuss the observations of successful cyber security operations from the virtual event showing the users a big picture about the organization how they successfully react to new specific accessories build on the cyber security.

    Provided By Carnegie Mellon University

  • Webcasts // Aug 2013

    Responding to a Large Scale Cybersecurity Incident

    In this Webcast, the presenter explain about how to respond to a large scale cyber security incidents, with an introduction about the role of technology. Technology which is an important one, but the limitations of that need to be understood.

    Provided By Carnegie Mellon University

  • Webcasts // Aug 2013

    20+ Years of Cyber (in)Security

    The use of information and communications technology is pervasive. The information systems are increasingly interconnected and interdependent. Attack technology is outpacing defensive technology. In this webcast, the presenter will give an introduction to the cyber security.

    Provided By Carnegie Mellon University

  • Webcasts // Aug 2013

    Achieving Mission Assurance Through Resilience Management

    In this webcast, the presenter going to explain about the resilience management. Resilience management is to protect a sustain organizations mission and business operations. The presenter will also explain about how the organizations are constant operational stress on these days.

    Provided By Carnegie Mellon University

  • Webcasts // Aug 2013

    Developing Your Cyber Workforce

    In this webcast, the presenter going to explain about the how to develop the cyber workforce against the security attacks. The presenter going to discuss about the new solutions and research that he had done to improve the security.

    Provided By Carnegie Mellon University

  • White Papers // Aug 2013

    A Proof of Correctness for Egalitarian Paxos

    In this paper the authors present a proof of correctness for Egalitarian Paxos (EPaxos), a new distributed consensus algorithm based on Paxos. EPaxos achieves three goals: availability without interruption as long as a simple majority of replicas are reachable - its availability is not interrupted when replicas crash or fail...

    Provided By Carnegie Mellon University

  • White Papers // Aug 2013

    Breaking the DDoS Attack Chain

    Department of Defense (DoD) communications and data networks continue to be targets for adversaries to deny operational use of those networks. Distributed Denial of Service (DDoS) is one such attack strategy that has proven to be an effective method of denying service to military, political, infrastructure, and economic targets. The...

    Provided By Carnegie Mellon University

  • Webcasts // Aug 2013

    The Cyber Security R&D Pipeline - Building Capability Through Science

    In this webcast, the presenter going to explain about the cyber security R&D pipeline how to build the capability through science. The presenter explains about the SEI advances software engineering and related disciplines to ensure systems with predictable and improved quality, cost and schedule.

    Provided By Carnegie Mellon University

  • White Papers // Jul 2013

    ReDABLS: Revisiting Device Attestation with Bounded Leakage of Secrets

    Many commodity operating systems and applications become infested with malicious software over time, primarily due to exploits that take advantage of software flaws and operator errors. In this paper, the authors present the salient features of a system design which allows remote-device authentication by a verifier, reaching malware-free system states,...

    Provided By Carnegie Mellon University

  • White Papers // Jun 2013

    A Case for Efficient Hardware/Software Cooperative Management of Storage and Memory

    Most applications manipulate persistent data, yet traditional systems decouple data manipulation from persistence in a two-level storage model. Programming languages and system software manipulate data in one set of formats in volatile main memory (DRAM) using a load/store interface, while storage systems maintain persistence in another set of formats in...

    Provided By Carnegie Mellon University

  • White Papers // Jun 2013

    Memory Scaling: A Systems Architecture Perspective

    The memory system is a fundamental performance and energy bottleneck in almost all computing systems. Recent system design, application, and technology trends that require more capacity, bandwidth, efficiency, and predictability out of the memory system make it an even more important system bottleneck. At the same time, DRAM technology is...

    Provided By Carnegie Mellon University

  • White Papers // Jun 2013

    Space-Efficient, High-Performance Rank & Select Structures on Uncompressed Bit Sequences

    Rank & select data structures are one of the fundamental building blocks for many modern succinct data structures. With the continued growth of massive-scale information services, the space efficiency of succinct data structures is becoming increasingly attractive in practice. In this paper, the authors re-examine the design of rank &...

    Provided By Carnegie Mellon University

  • White Papers // May 2013

    Resistance is Not Futile: Detecting DDoS Attacks without Packet Inspection

    Packets in anonymous networks are fully protected. Therefore, traditional methods relying on packet header and higher layer information do not work to detect Distributed-Denial-of-Service (DDoS) attacks in anonymous networks. In this paper, the authors propose to use observable statistics at routers that need no packet inspection to infer the presence...

    Provided By Carnegie Mellon University

  • White Papers // May 2013

    UnLocIn: Unauthorized Location Inference on Smartphones without Being Caught

    Location privacy has become one of the critical issues in the Smartphone era. Since users carry their phones everywhere and all the time, leaking users' location information can have dangerous implications. In this paper, the authors leverage the idea that Wi-Fi parameters not considered to be \"Sensitive\" in the Android...

    Provided By Carnegie Mellon University

  • White Papers // May 2013

    Network Self-Organization in the Internet of Things

    The Internet of Things is a paradigm that allows the interaction of ubiquitous devices through a network to achieve common goals. This paradigm like any man-made infrastructure is subject to disasters, outages and other adversarial conditions. Under these situations provisioned communications fail, rendering this paradigm with little or no use....

    Provided By Carnegie Mellon University

  • White Papers // May 2013

    How to Jam Without Getting Caught: Analysis and Empirical Study of Stealthy Periodic Jamming

    Despite the widespread commercial use of spread spectrum technology, advanced algorithms and modern hardware capabilities still allows efficient denial-of-service attacks against wireless communication systems using jamming. Much of the recent work on jamming mitigation has focused on how to adjust the transmitter-receiver system once a jamming attack has been detected....

    Provided By Carnegie Mellon University

  • White Papers // May 2013

    ASIA: Accelerated Secure In-network Aggregation in Vehicular Sensing Networks

    Vehicular Ad-Hoc NETworks (VANETs) can potentially become a sensing platform. In-network aggregation, a fundamental primitive for querying sensory data, has been shown to reduce overall communication overhead at large. To secure data aggregation in VANETs, existing schemes mainly rely on digital signatures. However, generating and verifying such signatures can cause...

    Provided By Carnegie Mellon University

  • White Papers // May 2013

    Stochastic Optimization of Flow-Jamming Attacks in Multichannel Wireless Networks

    An attacker can launch an efficient jamming attack to deny service to flows in wireless networks by using cross-layer knowledge of the target network. For example, flow-jamming defined in existing work incorporates network layer information into the conventional jamming attack to maximize its attack efficiency. In this paper, the authors...

    Provided By Carnegie Mellon University

  • White Papers // May 2013

    Wireless Mesh Network Simulator for Studying Cross-Layer Jamming Effects

    Various wireless mesh network standards have been actively constituted for the last several years. Because of its flexible network architecture, wireless mesh network can provide alternative paths even when some of wireless links are broken by node failures or intended attacks. Among various types of mesh network, the authors focus...

    Provided By Carnegie Mellon University

  • White Papers // May 2013

    MeshJam: Intelligent Jamming Attack and Defense in IEEE 802.11s Wireless Mesh Networks

    Wireless mesh networks represent an emerging network architecture which has been actively studied and standardized for the last several years. Because of their flexible network architecture, wireless mesh networks can provide alternative paths even when wireless links are broken by node failures or routing attacks. Among a variety of mesh...

    Provided By Carnegie Mellon University

  • White Papers // May 2013

    Specialized Storage for Big Numeric Time Series

    Big Data analysis is being used today to yield extraordinary insights in a variety of fields, including cancer research, traffic congestion, and datacenter health. Numeric time series data has unique storage requirements and access patterns that can benefit from specialized support, given its importance in big data analyses. Popular frameworks...

    Provided By Carnegie Mellon University

  • White Papers // Jan 2011

    Automation Without Predictability Is a Recipe for Failure

    Automated management seems a must, as distributed systems and datacenters continue to grow in scale and complexity. But, automation of performance problem diagnosis and tuning relies upon predictability, which in turn relies upon low variance - most automation tools aren't effective when variance is regularly high. This paper argues that,...

    Provided By Carnegie Mellon University

  • White Papers // May 2011

    Privacy-Sensitive VM Retrospection

    The success of cloud computing leads to large centralized collections of Virtual Machine (VM) images. The ability to retrospect (examine the historical state of) these images at a high semantic level can be valuable in many aspects of IT management such as debugging and troubleshooting, software quality control, legal establishment...

    Provided By Carnegie Mellon University

  • White Papers // Nov 2011

    Efficient Exploratory Testing of Concurrent Systems

    In the authors' experience, exploratory testing has reached a level of maturity that makes it a practical and often the most cost-effective approach to testing. Notably, previous paper has demonstrated that exploratory testing is capable of finding bugs even in well-tested systems. However, the number of bugs found gives little...

    Provided By Carnegie Mellon University

  • White Papers // Oct 2011

    The Case for Sleep States in Servers

    While sleep states have existed for mobile devices and workstations for some time, these sleep states have largely not been incorporated into the servers in today's data centers. Chip designers have been unmotivated to design sleep states because data center administrators haven't expressed any desire to have them. High setup...

    Provided By Carnegie Mellon University

  • White Papers // Nov 2011

    Active Disk Meets Flash: A Case for Intelligent SSDs

    The intelligent Solid-State Drives (iSSDs) allow execution of limited application functions (e.g., data filtering or aggregation) on their internal hardware resources, exploiting SSD characteristics and trends to provide large and growing performance and energy efficiency benefits. Most notably, internal flash media bandwidth can be significantly (2 - 4 or more)...

    Provided By Carnegie Mellon University

  • White Papers // Nov 2012

    Persistent, Protected and Cached: Building Blocks for Main Memory Data Stores

    In this paper the authors look at systems design for consistent, durable, and safe memory management for future byte-addressable Non-Volatile (NV) memory. Specifically, they focus on how application-level interfaces need to change to accommodate this memory on the main memory bus and propose a new NV-memory-aware memory allocator that incorporates...

    Provided By Carnegie Mellon University

  • White Papers // Feb 2014

    SpringFS: Bridging Agility and Performance in Elastic Distributed Storage

    The elastic storage systems can be expanded or contracted to meet current demand, allowing servers to be turned off or used for other tasks. However, the usefulness of an elastic distributed storage system is limited by its agility: how quickly it can increase or decrease its number of servers. Due...

    Provided By Carnegie Mellon University

  • White Papers // Dec 2013

    Tetrisched: Space-Time Scheduling for Heterogeneous Datacenters

    Tetrisched is a new scheduler that explicitly considers both job-specific preferences and estimated job runtimes in its allocation of resources. Combined, this information allows tetrisched to provide higher overall value to complex application mixes consolidated on heterogeneous collections of machines. Job-specific preferences, provided by tenants in the form of composable...

    Provided By Carnegie Mellon University

  • White Papers // Jan 2014

    Toward Strong, Usable Access Control for Shared Distributed Data

    As non-expert users produce increasing amounts of personal digital data, usable access control becomes critical. Current approaches often fail, because they insufficiently protect data or confuse users about policy specification. This paper presents penumbra, a distributed file system with access control designed to match users' mental models while providing principled...

    Provided By Carnegie Mellon University

  • White Papers // Nov 2013

    More Effective Distributed ML via a Stale Synchronous Parallel Parameter Server

    The authors propose a parameter server system for distributed ML, which follows a Stale Synchronous Parallel (SSP) model of computation that maximizes the time computational workers spend doing useful work on ML algorithms, while still providing correctness guarantees. The parameter server provides an easy-to-use shared interface for read/write access to...

    Provided By Carnegie Mellon University

  • White Papers // Sep 2013

    Program Interference in MLC NAND Flash Memory: Characterization, Modeling, and Mitigation

    As NAND flash memory continues to scale down to smaller process technology nodes, its reliability and endurance are degrading. One important source of reduced reliability is the phenomenon of program interference: when a flash cell is programmed to a value, the programming operation affects the threshold voltage of not only...

    Provided By Carnegie Mellon University

  • White Papers // Sep 2013

    LightTx: A Lightweight Transactional Design in Flash-based SSDs to Support Flexible Transactions

    Flash memory has accelerated the architectural evolution of storage systems with its unique characteristics compared to magnetic disks. The no-overwrite property of flash memory has been leveraged to efficiently support transactions, a commonly used mechanism in systems to provide consistency. However, existing transaction designs embedded in flash-based Solid State Drives...

    Provided By Carnegie Mellon University

  • White Papers // Jun 2013

    Memory Scaling: A Systems Architecture Perspective

    The memory system is a fundamental performance and energy bottleneck in almost all computing systems. Recent system design, application, and technology trends that require more capacity, bandwidth, efficiency, and predictability out of the memory system make it an even more important system bottleneck. At the same time, DRAM technology is...

    Provided By Carnegie Mellon University

  • White Papers // Oct 2013

    Challenges in Security and Privacy for Mobile Edge-Clouds

    Mobile devices such as Smartphone and tablets are ubiquitous today, and many of them possess significant computation power, powerful sensors such as high-resolution cameras and GPS sensors, and a wealth of sensor data such as photos, videos, and location information. Collections of mobile devices in close geographical proximity present both...

    Provided By Carnegie Mellon University

  • White Papers // Aug 2013

    A Proof of Correctness for Egalitarian Paxos

    In this paper the authors present a proof of correctness for Egalitarian Paxos (EPaxos), a new distributed consensus algorithm based on Paxos. EPaxos achieves three goals: availability without interruption as long as a simple majority of replicas are reachable - its availability is not interrupted when replicas crash or fail...

    Provided By Carnegie Mellon University

  • White Papers // Jun 2013

    A Case for Efficient Hardware/Software Cooperative Management of Storage and Memory

    Most applications manipulate persistent data, yet traditional systems decouple data manipulation from persistence in a two-level storage model. Programming languages and system software manipulate data in one set of formats in volatile main memory (DRAM) using a load/store interface, while storage systems maintain persistence in another set of formats in...

    Provided By Carnegie Mellon University

  • White Papers // Apr 2013

    Asymmetry-Aware Execution Placement on Manycore Chips

    Network-on-chip based manycore systems with multiple memory controllers on a chip are gaining prevalence. Among other research considerations, placing an increasing number of cores on a chip creates a type of resource access asymmetries that didn't exist before. A common assumption of uniform or hierarchical memory controller access no longer...

    Provided By Carnegie Mellon University

  • White Papers // May 2013

    Building a High-Performance Metadata Service by Reusing Scalable I/O Bandwidth

    Modern parallel and cluster file systems provide highly scalable I/O bandwidth by enabling highly parallel access to file data. Unfortunately metadata access does not benefit from parallel data transfer, so metadata performance scaling is less common. To support metadata-intensive workloads, the authors offer a middleware design that layers on top...

    Provided By Carnegie Mellon University

  • White Papers // Feb 2013

    Evaluating STT-RAM as an Energy-Efficient Main Memory Alternative

    In this paper, the authors explore the possibility of using STT-RAM technology to completely replace DRAM in main memory. Their goal is to make STT-RAM performance comparable to DRAM while providing substantial power savings. Towards this goal, they first analyze the performance and energy of STTRAM, and then identify key...

    Provided By Carnegie Mellon University

  • White Papers // May 2013

    Specialized Storage for Big Numeric Time Series

    Big Data analysis is being used today to yield extraordinary insights in a variety of fields, including cancer research, traffic congestion, and datacenter health. Numeric time series data has unique storage requirements and access patterns that can benefit from specialized support, given its importance in big data analyses. Popular frameworks...

    Provided By Carnegie Mellon University

  • White Papers // May 2012

    Concurrent Systematic Testing at Scale

    Systematic testing, first demonstrated in small, specialized cases 15 years ago, has matured sufficiently for large-scale systems developers to begin to put it into practice. With actual deployment comes new, pragmatic challenges to the usefulness of the techniques. In this paper the authors are concerned with scaling dynamic partial order...

    Provided By Carnegie Mellon University

  • White Papers // May 2012

    Distributed Directories using Giga+ and PVFS

    There is growing set of large-scale data-intensive applications that require file system directories to store millions to billions of files in each directory and to sustain hundreds of thousands of concurrent directory operations per second. Unfortunately, most cluster file systems are unable to provide this level of scale and parallelism....

    Provided By Carnegie Mellon University

  • White Papers // May 2012

    Shingled Magnetic Recording for Big Data Applications

    Modern Hard Disk Drives (HDDs) are fast approaching the super-paramagnetic limit forcing the storage industry to look for innovative ways to transition from traditional magnetic recording to heat-assisted magnetic recording or bit-patterned magnetic recording. Shingled Magnetic Recording (SMR) is a step in this direction as it delivers high storage capacity...

    Provided By Carnegie Mellon University

  • White Papers // Jun 2012

    Hadoop's Adolescence: A Comparative Workload Analysis from Three Research Clusters

    The authors analyze Hadoop workloads from three different research clusters from an application-level perspective, with two goals: explore new issues in application patterns and user behavior and understand key performance challenges related to IO and load balance. Their analysis suggests that Hadoop usage is still in its adolescence. They see...

    Provided By Carnegie Mellon University

  • White Papers // Aug 2012

    A Case for Small Row Buffers in Non-Volatile Main Memories

    DRAM-based main memories have read operations that destroy the read data, and as a result, must buffer large amounts of data on each array access to keep chip costs low. Unfortunately, system-level trends such as increased memory contention in multi-core architectures and data mapping schemes that improve memory parallelism lead...

    Provided By Carnegie Mellon University

  • White Papers // Aug 2012

    Row Buffer Locality Aware Caching Policies for Hybrid Memories

    Phase Change Memory (PCM) is a promising technology that can offer higher capacity than DRAM. Unfortunately, PCM's access latency and energy are higher than DRAM's and its endurance is lower. Many DRAM-PCM hybrid memory systems use DRAM as a cache to PCM, to achieve the low access latency and energy,...

    Provided By Carnegie Mellon University

  • White Papers // Oct 2012

    JackRabbit: Improved Agility in Elastic Distributed Storage

    Elastic storage systems can be expanded or contracted to meet current demand, allowing servers to be turned off or used for other tasks. However, the usefulness of an elastic distributed storage system is limited by its agility: how quickly it can increase or decrease its number of servers. This paper...

    Provided By Carnegie Mellon University

  • White Papers // Sep 2013

    HAT: Heterogeneous Adaptive Throttling for On-Chip Networks

    The Network-on-Chip (NoC) is a primary shared resource in a Chip Multi-Processor (CMP) system. As core counts continue to increase and applications become increasingly data-intensive, the network load will also increase, leading to more congestion in the network. This network congestion can degrade system performance if the network load is...

    Provided By Carnegie Mellon University

  • White Papers // Nov 2012

    MemC3: Compact and Concurrent MemCache with Dumber Caching and Smarter Hashing

    MemC3 is an in-memory key-value store that is designed to provide caching service for read-mostly workloads. It is built on carefully engineered algorithms and data structures with a set of architecture-aware and workload-ware optimizations to achieve high concurrency, space-efficiency and cache-locality. In particular, MemC3 uses a new hashing scheme -...

    Provided By Carnegie Mellon University

  • White Papers // Nov 2012

    HPC Computation on Hadoop Storage with PLFS

    In this paper the authors describe how they adapted the Parallel Log Structured File-System (PLFS) to enable HPC applications to be able read and write data from the HDFS cloud storage subsystem. Their enhanced version of PLFS provides HPC applications with the ability to concurrently write from multiple compute nodes...

    Provided By Carnegie Mellon University

  • White Papers // Jan 2013

    TABLEFS: Enhancing Metadata Efficiency in the Local File System

    File systems that manage magnetic disks have long recognized the importance of sequential allocation and large transfer sizes for file data. Fast random access has dominated metadata lookup data structures with increasing use of B-trees on-disk. Yet the authors' experiments with workloads dominated by metadata and small file access indicate...

    Provided By Carnegie Mellon University

  • White Papers // Jan 2013

    Giga+TableFS on PanFS: Scaling Metadata Performance on Cluster File Systems

    Modern file systems provide scalable performance for large file data management. However, in case of metadata management the usual approach is to have single or few points of Meta-Data Service (MDS). In the current world, file systems are challenged by unique needs such as managing exponentially growing files, using file-system...

    Provided By Carnegie Mellon University

  • White Papers // Apr 2013

    Tiered-Latency DRAM: A Low Latency and Low Cost DRAM Architecture

    The capacity and cost-per-bit of DRAM have historically scaled to satisfy the needs of increasingly large and complex computer systems. However, DRAM latency has remained almost constant, making memory latency the performance bottleneck in today's systems. The authors observe that the high access latency is not intrinsic to DRAM, but...

    Provided By Carnegie Mellon University

  • White Papers // Apr 2013

    Practical Batch-Updatable External Hashing with Sorting

    In this paper the authors present a practical external hashing scheme that supports fast lookup (7 microseconds) for large datasets (millions to billions of items) with a small memory footprint (2.5 bits/item) and fast index construction (151 K items/s for 1-KiB key-value pairs). Their scheme combines three key techniques: a...

    Provided By Carnegie Mellon University

  • White Papers // Jan 2013

    Application-to-Core Mapping Policies to Reduce Memory System Interference in Multi-Core Systems

    Future manycore processors are likely to concurrently execute a large number of diverse applications. How these applications are mapped to cores largely determines the interference between these applications in critical shared hardware resources. This paper proposes new application-to-core mapping policies to improve system performance by reducing inter-application interference in the...

    Provided By Carnegie Mellon University

  • White Papers // Jan 2013

    MISE: Providing Performance Predictability and Improving Fairness in Shared Main Memory Systems

    Applications running concurrently on a multi-core system interfere with each other at the main memory. This interference can slow down different applications differently. Accurately estimating the slowdown of each application in such a system can enable mechanisms that can enforce quality-of-service. While much prior work has focused on mitigating the...

    Provided By Carnegie Mellon University

  • White Papers // Jan 2012

    ZZFS: A Hybrid Device and Cloud File System for Spontaneous Users

    A good execution of data placement, caching and consistency policies across a user's personal devices has always been hard. Unpredictable networks, capricious user behavior with leaving devices on or off and non-uniform energy-saving policies constantly interfere with the good intentions of a storage system's policies. This paper's contribution is to...

    Provided By Carnegie Mellon University

  • White Papers // Feb 2012

    Near-Real-Time Inference of File-Level Mutations from Virtual Disk Writes

    The authors describe a new mechanism for cloud computing enabling near-real-time monitoring of virtual disk write streams across an entire cloud. Their solution has low IO overhead for the guest VM, low latency to file-level mutation notification, and a layered design for scalability. They achieve low IO overhead by duplicating...

    Provided By Carnegie Mellon University

  • White Papers // May 2012

    TABLEFS: Embedding a NoSQL Database Inside the Local File System

    Conventional file systems are optimized for large file transfers instead of workloads that are dominated by metadata and small file accesses. This paper examines using techniques adopted from NoSQL databases to manage file system metadata and small files, which feature high rate of changes and efficient out-of-core data representation. A...

    Provided By Carnegie Mellon University

  • White Papers // Apr 2012

    Towards Understanding Heterogeneous Clouds at Scale: Google Trace Analysis

    With the emergence of large, heterogeneous, shared computing clusters, their efficient use by mixed distributed workloads and tenants remains an important challenge. Unfortunately, little data has been available about such workloads and clusters. This paper analyzes a recent Google release of scheduler request and utilization data across a large (12500+)...

    Provided By Carnegie Mellon University