In an effort to increase the utilization of data center resources cloud providers have introduced a new type of virtual machine (VM) offering, called a burstable VM (BVM). Our work is the first to study the characteristics of burstable VMs (based on traces from production systems at a major cloud provider) and resource allocation approaches for BVM workloads. We propose new approaches for BVM resource allocation and use extensive simulations driven by field data to compare them with two baseline approaches used in practice. We find that traditional approaches based on using a fixed oversubscription ratio or based on the Central Limit Theorem do not work well for BVMs: They lead to either low utilization or high server capacity violation rates. Based on the lessons learned from our workload study, we develop a new approach to BVM scheduling, called Audible, using a non-parametric statistical model, which makes the approach light-weight and workload independent, and obviates the need for training machine learning models and for tuning their parameters. We show that Audible achieves high system utilization while being able to enforce stringent requirements on server capacity violations.

High density Solid State Drives, such as QLC drives, offer increased storage capacity, but a magnitude lower Program and Erase (P/E) cycles, limiting their endurance and hence usability. We present the design and implementation of non-binary, Voltage-Based Write-Once-Memory (WOM-v) Codes to improve the lifetime of QLC drives. First, we develop a FEMU based simulator test-bed to evaluate the gains of WOM-v codes on real world workloads. Second, we propose and implement two optimizations, an efficient garbage collection mechanism and an encoding optimization to drastically improve WOM-v code endurance without compromising performance. Third, we propose analytical approaches to obtain estimates of the endurance gains under WOM-v codes. We analyze the Greedy garbage collection technique with uniform page access distribution and the Least Recently Written (LRW) garbage collection technique with skewed page access distribution in the context of WOM-v codes. We find that although both approaches overestimate the number of required erase operations, the model based on greedy garbage collection with uniform page access distribution provides tighter bounds. A careful evaluation, including microbenchmarks and trace-driven evaluation, demonstrates that WOM-v codes can reduce Erase cycles for QLC drives by 4.4×–11.1× for real world workloads with minimal performance overheads resulting in improved QLC SSD lifetime.

This article presents the first large-scale field study of NAND-based SSDs in enterprise storage systems (in contrast to drives in distributed data center storage systems). The study is based on a very comprehensive set of field data, covering 1.6 million SSDs of a major storage vendor (NetApp). The drives comprise three different manufacturers, 18 different models, 12 different capacities, and all major flash technologies (SLC, cMLC, eMLC, 3D-TLC). The data allows us to study a large number of factors that were not studied in prior works, including the effect of firmware versions, the reliability of TLC NAND, and the correlations between drives within a RAID system. This article presents our analysis, along with a number of practical implications derived from it.

Storage systems are designed and optimized relying on wisdom derived from analysis studies of file-system and block-level workloads. However, while SSDs are becoming a dominant building block in many storage systems, their design continues to build on knowledge derived from analysis targeted at hard disk optimization. Though still valuable, it does not cover important aspects relevant for SSD performance. In a sense, we are “searching under the streetlight,” possibly missing important opportunities for optimizing storage system design. We present the first I/O workload analysis designed with SSDs in mind. We characterize traces from four repositories and examine their “temperature” ranges, sensitivity to page size, and “logical locality.” We then take the first step towards correlating these characteristics with three standard performance metrics: write amplification, read amplification, and flash read costs. Our results show that SSD-specific characteristics strongly affect performance, often in surprising ways.

As solid state drives (SSDs) are increasingly replacing hard disk drives, the reliability of storage systems depends on the failure modes of SSDs and the ability of the file system layered on top to handle these failure modes. While the classical paper on IRON File Systems provides a thorough study of the failure policies of three file systems common at the time, we argue that 13 years later it is time to revisit file system reliability with SSDs and their reliability characteristics in mind, based on modern file systems that incorporate journaling, copy-on-write, and log-structured approaches and are optimized for flash. This article presents a detailed study, spanning ext4, Btrfs, and F2FS, and covering a number of different SSD error modes. We develop our own fault injection framework and explore over 1,000 error cases. Our results indicate that 16% of these cases result in a file system that cannot be mounted or even repaired by its system checker. We also identify the key file system metadata structures that can cause such failures, and, finally, we recommend some design guidelines for file systems that are deployed on top of SSDs.

As the scale of High-Performance Computing (HPC) clusters continues to grow, their increasing failure rates and energy consumption levels are emerging as serious design concerns. Efficiently running systems at such large scales critically relies on deploying effective, practical methods for fault tolerance while having a good understanding of their respective performance and energy overheads. The most commonly used fault tolerance method is checkpoint/restart. Checkpoint scheduling policies, however, have been traditionally optimized and analysed from one angle: application performance. In this work, we provide an extensive analysis of the performance, energy and I/O costs associated with a wide array of checkpointing policies. We consider practical deployment issues and show that simple formulas can be used to accurately estimate wasted work in a system. We propose methods to optimize checkpoint scheduling for energy savings and evaluate the runtime-optimized and energy-optimized policies using simulations based on failure logs from 10 production HPC clusters. Our results show ample room for achieving high quality energy/performance tradeoffs when using methods that exploit characteristics of real world failures. We also analyze the impact of energy-optimized checkpointing on the storage subsystem and identify policies that are optimal for I/O savings.

In large-scale computing platforms, jobs are prone to interruptions and premature terminations, limiting their usability and leading to significant waste in cluster resources. In this paper, we tackle this problem in three steps. First, we provide a comprehensive study based on log data from multiple large-scale production systems to identify patterns in the behaviour of unsuccessful jobs across different clusters and investigate possible root causes behind job termination. Our results reveal several interesting properties that distinguish unsuccessful jobs from others, particularly w.r.t. resource consumption patterns and job configuration settings. Secondly, we design a machine learning-based framework for predicting job and task terminations. We show that job failures can be predicted relatively early with high precision and recall, and also identify attributes that have strong predictive power of job failure. Finally, we demonstrate in a concrete use case how our prediction framework can be used to mitigate the effect of unsuccessful execution using an effective task-cloning policy that we propose.

Solid-state drives (SSDs) based on NAND flash are making deep inroads into data centers as well as the consumer market. In 2016, manufacturers shipped more than 130 million units totaling around 50 Exabytes of storage capacity. As the amount of data stored on solid state drives keeps increasing, it is important to understand the reliability characteristics of these devices. For a long time, our knowledge about flash reliability was derived from controlled experiments in lab environments under synthetic workloads, often using methods for accelerated testing. However, within the last two years, three large-scale field studies have been published that report on the failure behavior of flash devices in production environments subjected to real workloads and operating conditions. The goal of this paper is to provide an overview of what we have learned about flash reliability in production, and where appropriate contrasting it with prior studies performing controlled experiments.

In a data center, an IO from an application to distributed storage traverses not only the network but also several software stages with diverse functionality. This set of ordered stages is known as the storage or IO stack. Stages include caches, hypervisors, IO schedulers, file systems, and device drivers. Indeed, in a typical data center, the number of these stages is often larger than the number of network hops to the destination. Yet, while packet routing is fundamental to networks, no notion of IO routing exists on the storage stack. The path of an IO to an endpoint is predetermined and hard coded. This forces IO with different needs (e.g., requiring different caching or replica selection) to flow through a one-size-fits-all IO stack structure, resulting in an ossified IO stack. This article proposes sRoute, an architecture that provides a routing abstraction for the storage stack. sRoute comprises a centralized control plane and “sSwitches” on the data plane. The control plane sets the forwarding rules in each sSwitch to route IO requests at runtime based on application-specific policies. A key strength of our architecture is that it works with unmodified applications and Virtual Machines (VMs). This article shows significant benefits of customized IO routing to data center tenants: for example, a factor of 10 for tail IO latency, more than 60% better throughput for a customized replication protocol, a factor of 2 in throughput for customized caching, and enabling live performance debugging in a running system.

At the core of the "Big Data" revolution lie frameworks and systems that allow for the massively parallel processing of large amounts of data. Ironically, while they have been designed for processing large amounts of data, these systems are at the same time major producers of data: to support the administration and management of these huge-scale systems, they are configured to generate detailed log and monitoring data, periodically capturing the system state across all nodes, components and jobs in the system. While such logging information is used routinely by sysadmins for ad-hoc trouble-shooting and problem diagnosis, we point out that there is a tremendous value in analyzing such data from a research point of view. In this talk, we will go over several case studies that demonstrate how measuring and analyzing measurement data from production systems can provide new insights into how systems work and fail, and how these new insights can help in designing better systems.

In data centers, caches work both to provide low IO latencies and to reduce the load on the back-end network and storage. But they are not designed for multi-tenancy; system-level caches today cannot be configured to match tenant or provider objectives. Exacerbating the problem is the increasing number of un-coordinated caches on the IO data plane. The lack of global visibility on the control plane to coordinate this distributed set of caches leads to inefficiencies, increasing cloud provider cost.

Efficient use of high-performance computing (HPC) installations critically relies on effective methods for fault tolerance. The most commonly used method is checkpoint/restart, where an application writes periodic checkpoints of its state to stable storage that it can restart from in the case of a failure. Despite the prevalence of checkpoint/restart, it is still not very well understood in practice how to set its key parameter, the checkpoint interval. Despite a large body of theoretical work, practitioners still rely on crude rules-of-thumb such as “checkpoint once every hour”. Our goal is to identify methods for optimizing the checkpointing process that are easy to use in practice and at the same time achieve high quality solutions. In particular, our paper makes the following contributions: We evaluate an array of methods for optimizing the checkpoint interval, some previously known as well as new ones that we propose, using real-world failure logs. We show that a very simple closed-form solution can easily be adapted for use in practice and achieves near-optimal performance. We also find that more complex solutions only negligibly improve performance based on real world traces. We show that simple back-of-the envelope formulas can be used to accurately estimate the wasted work in HPC systems, and make projections of future HPC systems and requirements for their efficient use.

As the scale of high-performance computing (HPC) clusters continues to grow, their increasing failure rates and energy consumption levels are emerging as two serious design concerns that are expected to become more challenging in future Exascale systems. Therefore, efficiently running systems at such large scales requires an in-depth understanding of the performance and energy costs associated with different fault tolerance techniques. The most commonly used fault tolerance method is checkpoint/restart. Over the years, checkpoint scheduling policies have been traditionally optimized and analysed from a performance perspective. Understanding the energy profile of these policies or how to optimize them for energy savings (rather than performance), remain not very well understood. In this paper, we provide an extensive analysis of the energy/ performance tradeoffs associated with an array of checkpoint scheduling policies, including policies that we propose, as well as few existing ones in the literature. We estimate the energy overhead for a given checkpointing policy, and provide simple formulas to optimize checkpoint scheduling for energy savings, with or without a bound on runtime. We then evaluate and compare the runtime-optimized and energy-optimized versions of the different methods using trace driven simulations based on failure logs from 10 production HPC clusters. Our results show ample room for achieving high energy savings with a low runtime overhead when using non-constant (adaptive) checkpointing methods that exploit characteristics of HPC failures. We also analyze the impact of energy-optimized checkpointing on the storage subsystem, identify policies that are more optimal for I/O savings, and study how to optimize for energy with a bound on I/O time.

The reliability and availability of today's large-scale systems hinges on the reliability of the often millions of hardware components they comprise. Before deployment, devices undergo rigorous testing as part of the design and manufacturing process to assure they meet reliability expectations. In this talk we will look at the other half of the story: the post-deployment life of devices, once they enter production use in the field. Based on field data from large-scale production systems, we will study different aspects of hardware reliability in the wild, with a focus on DRAM DIMMs, and show that life in in the real world can be quite different from that in the lab.

No abstract available.

As the component count in supercomputing installations continues to increase, system reliability is becoming one of the major issues in designing HPC systems. These issues will become more challenging in future Exascale systems, which are predicted to include millions of CPU cores. Even with relatively reliable individual components, the sheer number of components will increase failure rates to unprecedented levels. Efficiently running those systems will require a good understanding of how different factors impact system reliability. In this paper we use a decade worth of field data made available by Los Alamos National Lab to study the impact of a diverse set of factors on the reliability of HPC systems. We provide insights into the nature of correlations between failures, and investigate the impact of factors, such as the power quality, temperature, fan and chiller reliability, system usage and utilization, and external factors, such as cosmic radiation, on system reliability.

Main memory is one of the leading hardware causes for machine crashes in today's datacenters. Designing, evaluating and modeling systems that are resilient against memory errors requires a good understanding of the underlying characteristics of errors in DRAM in the field. While there have recently been a few first studies on DRAM errors in production systems, these have been too limited in either the size of the data set or the granularity of the data to conclusively answer many of the open questions on DRAM errors. Such questions include, for example, the prevalence of soft errors compared to hard errors, or the analysis of typical patterns of hard errors. In this paper, we study data on DRAM errors collected on a diverse range of production systems in total covering nearly 300 terabyte-years of main memory. As a first contribution, we provide a detailed analytical study of DRAM error characteristics, including both hard and soft errors. We find that a large fraction of DRAM errors in the field can be attributed to hard errors and we provide a detailed analytical study of their characteristics. As a second contribution, the paper uses the results from the measurement study to identify a number of promising directions for designing more resilient systems and evaluates the potential of different protection mechanisms in the light of realistic error patterns. One of our findings is that simple page retirement policies might be able to mask a large number of DRAM errors in production systems, while sacrificing only a negligible fraction of the total DRAM in the system.

Latent sector errors (LSEs) are a common hard disk failure mode, where disk sectors become inaccessible while the rest of the disk remains unaffected. To protect against LSEs, commercial storage systems use scrubbers: background processes verifying disk data. The efficiency of different scrubbing algorithms in detecting LSEs has been studied in depth; however, no attempts have been made to evaluate or mitigate the impact of scrubbing on application performance. We provide the first known evaluation of the performance impact of different scrubbing policies in implementation, including guidelines on implementing a scrubber. To lessen this impact, we present an approach giving conclusive answers to the questions: when should scrubbing requests be issued, and at what size, to minimize impact and maximize scrubbing throughput for a given workload. Our approach achieves six times more throughput, and up to three orders of magnitude less slowdown than the default Linux I/O scheduler.

The energy consumed by data centers is starting to make up a significant fraction of the world's energy consumption and carbon emissions. A large fraction of the consumed energy is spent on data center cooling, which has motivated a large body of work on temperature management in data centers. Interestingly, a key aspect of temperature management has not been well understood: controlling the setpoint temperature at which to run a data center's cooling system. Most data centers set their thermostat based on (conservative) suggestions by manufacturers, as there is limited understanding of how higher temperatures will affect the system. At the same time, studies suggest that increasing the temperature setpoint by just one degree could save 2-5% of the energy consumption. This paper provides a multi-faceted study of temperature management in data centers. We use a large collection of field data from different production environments to study the impact of temperature on hardware reliability, including the reliability of the storage subsystem, the memory subsystem and server reliability as a whole. We also use an experimental testbed based on a thermal chamber and a large array of benchmarks to study two other potential issues with higher data center temperatures: the effect on server performance and power. Based on our findings, we make recommendations for temperature management in data centers, that create the potential for saving energy, while limiting negative effects on system reliability and performance.

Designing highly dependable systems requires a good understanding of failure characteristics. Unfortunately, little raw data on failures in large IT installations are publicly available. This paper analyzes failure data collected at two large high-performance computing sites. The first data set has been collected over the past nine years at Los Alamos National Laboratory (LANL) and has recently been made publicly available. It covers 23,000 failures recorded on more than 20 different systems at LANL, mostly large clusters of SMP and NUMA nodes. The second data set has been collected over the period of one year on one large supercomputing system comprising 20 nodes and more than 10,000 processors. We study the statistics of the data, including the root cause of failures, the mean time between failures, and the mean time to repair. We find, for example, that average failure rates differ wildly across systems, ranging from 20-1000 failures per year, and that time between failures is modeled well by a Weibull distribution with decreasing hazard rate. From one system to another, mean repair time varies from less than an hour to more than a day, and repair times are well modeled by a lognormal distribution.

Latent sector errors (LSEs) refer to the situation where particular sectors on a drive become inaccessible. LSEs are a critical factor in data reliability, since a single LSE can lead to data loss when encountered during RAID reconstruction after a disk failure or in systems without redundancy. LSEs happen at a significant rate in the field [Bairavasundaram et al. 2007], and are expected to grow more frequent with new drive technologies and increasing drive capacities. While two approaches, data scrubbing and intra-disk redundancy, have been proposed to reduce data loss due to LSEs, none of these approaches has been evaluated on real field data. This article makes two contributions. We provide an extended statistical analysis of latent sector errors in the field, specifically from the view point of how to protect against LSEs. In addition to providing interesting insights into LSEs, we hope the results (including parameters for models we fit to the data) will help researchers and practitioners without access to data in driving their simulations or analysis of LSEs. Our second contribution is an evaluation of five different scrubbing policies and five different intra-disk redundancy schemes and their potential in protecting against LSEs. Our study includes schemes and policies that have been suggested before, but have never been evaluated on field data, as well as new policies that we propose based on our analysis of LSEs in the field.

An important threat to reliable storage of data is silent data corruption. In order to develop suitable protection mechanisms against data corruption, it is essential to understand its characteristics. In this article, we present the first large-scale study of data corruption. We analyze corruption instances recorded in production storage systems containing a total of 1.53 million disk drives, over a period of 41 months. We study three classes of corruption: checksum mismatches, identity discrepancies, and parity inconsistencies. We focus on checksum mismatches since they occur the most. We find more than 400,000 instances of checksum mismatches over the 41-month period. We find many interesting trends among these instances, including: (i) nearline disks (and their adapters) develop checksum mismatches an order of magnitude more often than enterprise-class disk drives, (ii) checksum mismatches within the same disk are not independent events and they show high spatial and temporal locality, and (iii) checksum mismatches across different disks in the same storage system are not independent. We use our observations to derive lessons for corruption-proof system design.

Component failure in large-scale IT installations is becoming an ever-larger problem as the number of components in a single cluster approaches a million. This article is an extension of our previous study on disk failures [Schroeder and Gibson 2007] and presents and analyzes field-gathered disk replacement data from a number of large production systems, including high-performance computing sites and internet services sites. More than 110,000 disks are covered by this data, some for an entire lifetime of five years. The data includes drives with SCSI and FC, as well as SATA interfaces. The mean time-to-failure (MTTF) of those drives, as specified in their datasheets, ranges from 1,000,000 to 1,500,000 hours, suggesting a nominal annual failure rate of at most 0.88%. We find that in the field, annual disk replacement rates typically exceed 1%, with 2--4% common and up to 13% observed on some systems. This suggests that field replacement is a fairly different process than one might predict based on datasheet MTTF. We also find evidence, based on records of disk replacements in the field, that failure rate is not constant with age, and that rather than a significant infant mortality effect, we see a significant early onset of wear-out degradation. In other words, the replacement rates in our data grew constantly with age, an effect often assumed not to set in until after a nominal lifetime of 5 years. Interestingly, we observe little difference in replacement rates between SCSI, FC, and SATA drives, potentially an indication that disk-independent factors such as operating conditions affect replacement rates more than component-specific ones. On the other hand, we see only one instance of a customer rejecting an entire population of disks as a bad batch, in this case because of media error rates, and this instance involved SATA disks. Time between replacement, a proxy for time between failure, is not well modeled by an exponential distribution and exhibits significant levels of correlation, including autocorrelation and long-range dependence.

Designing highly dependable systems requires a good understanding of failure characteristics. Unfortunately, little raw data on failures in large IT installations is publicly available. This paper analyzes failure data recently made publicly available by one of the largest high-performance computing sites. The data has been collected over the past 9 years at Los Alamos National Laboratory and includes 23000 failures recorded on more than 20 different systems, mostly large clusters of SMP and NUMA nodes. We study the statistics of the data, including the root cause of failures, the mean time between failures, and the mean time to repair. We find for example that average failure rates differ wildly across systems, ranging from 20-1000 failures per year, and that time between failures is modeled well by a Weibull distribution with decreasing hazard rate. From one system to another, mean repair time varies from less than an hour to more than a day, and repair times are well modeled by a lognormal distribution

Transaction processing systems lie at the core of modern e-commerce applications such as on-line retail stores, banks and airline reservation systems. The economic success of these applications depends on the ability to achieve high user satisfaction, since a single mouse-click is all that it takes a frustrated user to switch to a competitor. Given that system resources are limited and demands are varying, it is difficult to provide optimal performance to all users at all times. However, often transactions can be divided into different classes based on how important they are to the online retailer. For example, transactions initiated by a "big spending" client are more important than transactions from a client that only browses the site. A natural goal then is to ensure short delays for the class of important transactions, while for the less important transactions longer delays are acceptable.

Scheduling/prioritization of DBMS transactions is important for many applications that rely on database backends. A convenient way to achieve scheduling is to limit the number of transactions within the database, maintaining most of the transactions in an external queue, which can be ordered as desired by the application. While external scheduling has many advantages in that it doesn’t require changes to internal resources, it is also difficult to get right in that its performance depends critically on the particular multiprogramming limit used (the MPL), i.e. the number of transactions allowed into the database. If the MPL is too low, throughput will suffer, since not all DBMS resources will be utilized. On the other hand, if the MPL is too high, there is insufficient control on scheduling. The question of how to adjust theMPL to achieve both goals simultaneously is an open problem, not just for databases but in system design in general. Herein we study this problem in the context of transactional workloads, both via extensive experimentation and queueing theoretic analysis. We find that the two most critical factors in adjusting the MPL are the number of resources that the workload utilizes and the variability of the transactions’ service demands. We develop a feedback based controller, augmented by queueing theoretic models for automatically adjusting the MPL. Finally, we apply our methods to the specific problem of external prioritization of transactions. We find that external prioritization can be nearly as effective as internal prioritization, without any negative consequences, when the MPL is set appropriately.

With the large component count of coming PetaFLOPS systems, component failures will be frequent. The success of those systems will therefore depend on their ability to avoid, cope and recover from failures. Unfortunately, research in this area is hampered by the limited understanding of what failures in real, large-scale systems look like, due to lack of publicly available failure data. This paper introduces the Computer Failure Data Repository (CFDR), an effort being undertaken at CMU to collect, share, and analyze failure data. The CFDR has been stimulated by the recent public release of a large set of failure data that has been collected over the past 9 years at Los Alamos National Laboratory and includes 23000 failures on more than 20 different systems, mostly large clusters. We present a high-level overview of some of our results from analyzing this data and describe the ongoing efforts and long-term plans of the CFDR.

This article provides a detailed implementation study on the behavior of web serves that serve static requests where the load fluctuates over time (transient overload). Various external factors are considered, including WAN delays and losses and different client behavior models. We find that performance can be dramatically improved via a kernel-level modification to the web server to change the scheduling policy at the server from the standard FAIR (processor-sharing) scheduling to SRPT (shortest-remaining-processing-time) scheduling. We find that SRPT scheduling induces no penalties. In particular, throughput is not sacrificed and requests for long files experience only negligibly higher response times under SRPT than they did under the original FAIR scheduling.

OLTP and transactional workloads are increasingly common in computer systems, ranging from e-commerce to warehousing to inventory management. It is valuable to provide priority scheduling in these systems, to reduce the response time for the most important clients, e.g. the "big spenders". Two-phase locking, commonly used in DBMS, makes prioritization difficult, as transactions wait for locks held by others regardless of priority. Common lock scheduling solutions, including non-preemptive priority inheritance and preemptive abort, have performance drawbacks for TPC-C type workloads. The contributions of this paper are two-fold: (i) We provide a detailed statistical analysis of locking in TPC-C workloads with priorities under several common preemptive and non-preemptive lock prioritization policies. We determine why non-preemptive policies fail to sufficiently help high-priority transactions, and why preemptive policies excessively hurt low-priority transactions, (ii) We propose and implement a policy, POW, that provides all the benefits of preemptive prioritization without its penalties.

Transactional workloads are a hallmark of modern OLTP and Web applications, ranging from electronic commerce and banking to online shopping. Often, the database at the core of these applications is the performance bottleneck. Given the limited resources available to the database, transaction execution times can vary wildly as they compete and wait for critical resources. As the competitor is "only a click away", valuable (high-priority) users must be ensured consistently good performance via QoS and transaction prioritization. This paper analyzes and proposes prioritization for transactional workloads in traditional database systems (DBMS). This work first performs a detailed bottleneck analysis of resource usage by transactional workloads on commercial and noncommercial DBMS (IBM DB2, Post-greSQL, Shore) under a range of configurations. Second, this work implements and evaluates the performance of several preemptive and nonpreemptive DBMS prioritization policies in PostgreSQL and Shore. The primary contributions of this work include (i) understanding the bottleneck resources in transactional DBMS workloads and (ii) a demonstration that prioritization in traditional DBMS can provide 2x-5x improvement for high-priority transactions using simple scheduling policies, without expense to low-priority transactions.

Is it possible to reduce the expected response time of every request at a web server, simply by changing the order in which we schedule the requests? That is the question we ask in this paper.This paper proposes a method for improving the performance of web servers servicing static HTTP requests. The idea is to give preference to requests for small files or requests with short remaining file size, in accordance with the SRPT (Shortest Remaining Processing Time) scheduling policy.The implementation is at the kernel level and involves controlling the order in which socket buffers are drained into the network. Experiments are executed both in a LAN and a WAN environment. We use the Linux operating system and the Apache and Flash web servers.Results indicate that SRPT-based scheduling of connections yields significant reductions in delay at the web server. These result in a substantial reduction in mean response time and mean slowdown for both the LAN and WAN environments. Significantly, and counter to intuition, the requests for large files are only negligibly penalized or not at all penalized as a result of SRPT-based scheduling.

While the MPP is still the most common architecture in supercomputer centers today, a simpler and cheaper machine configuration is appearing at many supercomputing sites. This alternative setup may be described simply as a collection of multiprocessors or a distributed server system. This collection of multiprocessors is fed by a single common stream of jobs, where each job is dispatched to exactly one of the multiprocessor machines for processing. The biggest question which arises in such distributed server systems is what is a good rule for assigning jobs to host machines: i.e. what is a good task assignment policy. Many task assignment policies have been proposed, but not systematically evaluated under supercomputing workloads. We start by comparing existing task assignment policies using a trace-driven simulation under supercomputing workloads. We validate our experiments by providing analytical proofs of the performance of each of these policies. These proofs also help provide much intuition. We find that while the performance of supercomputing servers varies widely with the task assignment policy, none of the above task assignment policies perform as well as we would like.