This document summarizes a presentation on transcendent memory (tmem) on the Xen virtual machine monitor. Tmem aims to solve the challenges of physical memory management in virtualized systems by creating a shared memory pool from unused and wasted guest memory. The presentation provides background on memory management in operating systems and virtual machine monitors, an overview of the tmem design and implementation progress, and a performance analysis comparing self-ballooning, tmem, tmem with deduplication, and tmem with deduplication and compression.
The document is a presentation on Transcendent Memory on Xen given at the Xen Summit in 2009. It discusses the challenges of optimizing physical memory distribution across virtual machines on a hypervisor. It introduces Transcendent Memory as a new approach that collects unused and wasted guest memory into a shared pool to better optimize memory allocation over time without performance penalties.
The document discusses energy-efficient storage in virtual machine environments. It notes that energy management is challenging due to the separation between the virtual machine monitor (VMM) and guest operating systems (OSs). Two approaches are proposed: early flush notifications from the VMM to VMs to synchronize buffer flushes, and buffering writes from VMs in the VMM when the disk is asleep to extend idle time. Evaluation shows these approaches reduce energy consumption by up to 14.8% compared to standard disk management in environments with multiple VMs.
IBM Power Systems can leverage several IBM storage solutions to improve performance and flexibility for IBM i workloads. These include direct attachment of DS5000 storage arrays as well as virtualizing storage through PowerVM VIOS configurations using DS5000, SVC, XIV, and other storage arrays. Using SSDs in Power Systems can also provide major performance boosts for applications with random read-intensive workloads.
This document provides an overview of Hyper-V Dynamic Memory in Windows Server 2008 R2 SP1. It discusses how Dynamic Memory allows virtual machines to dynamically adjust their memory allocation based on usage to improve consolidation ratios with minimal performance impact. It covers the architecture including adding and removing memory from VMs, and how Dynamic Memory impacts root memory reserves and NUMA memory management. Alternative memory techniques from VMware like page sharing and second level paging are also briefly discussed.
This document discusses Hyper-V Dynamic Memory, a new feature in Windows Server 2008 R2 SP1 that allows virtual machines to dynamically adjust their memory usage. It aims to improve VM consolidation ratios with minimal performance impact. Dynamic Memory architecture is explained, including how memory is added and removed from VMs. The impact on root memory reserves and NUMA memory management is covered. Alternative memory techniques from VMware are also discussed.
The document discusses solid state drives (SSDs) and their benefits over traditional hard disk drives (HDDs). SSDs offer much faster read/write speeds and lower power consumption compared to HDDs. A mixed configuration using both SSDs and HDDs is typically recommended, with "hot" frequently accessed data stored on SSDs and "cold" infrequently accessed data stored on HDDs. The document provides examples of workloads and applications that are good fits for SSDs, and also discusses technologies used in SSDs like single-level cells and multi-level cells. It also covers how to identify and move hot data to SSDs for the AIX and IBM i operating systems.
This document summarizes a presentation on datacenter computing trends and problems. It discusses how cooling is a major source of energy inefficiency in datacenters. It also explains how servers are rarely fully utilized but operate least efficiently during common usage of 30% load. The document advocates for achieving better energy proportionality so servers can be more efficient during typical usage levels. It presents approaches like disaggregated memory and servers that break CPU-memory co-location to improve efficiency and consolidation.
The Dell Latitude D610 is a durable and flexible notebook designed for mobile professionals. Weighing less than 5 pounds, it delivers powerful performance from an Intel Pentium M processor while withstanding tough conditions through its magnesium alloy frame and sealed keyboard. It provides connectivity and expandability through features like wireless LAN, Gigabit Ethernet, USB ports, and an optional docking station.
The document is a presentation on Transcendent Memory on Xen given at the Xen Summit in 2009. It discusses the challenges of optimizing physical memory distribution across virtual machines on a hypervisor. It introduces Transcendent Memory as a new approach that collects unused and wasted guest memory into a shared pool to better optimize memory allocation over time without performance penalties.
The document discusses energy-efficient storage in virtual machine environments. It notes that energy management is challenging due to the separation between the virtual machine monitor (VMM) and guest operating systems (OSs). Two approaches are proposed: early flush notifications from the VMM to VMs to synchronize buffer flushes, and buffering writes from VMs in the VMM when the disk is asleep to extend idle time. Evaluation shows these approaches reduce energy consumption by up to 14.8% compared to standard disk management in environments with multiple VMs.
IBM Power Systems can leverage several IBM storage solutions to improve performance and flexibility for IBM i workloads. These include direct attachment of DS5000 storage arrays as well as virtualizing storage through PowerVM VIOS configurations using DS5000, SVC, XIV, and other storage arrays. Using SSDs in Power Systems can also provide major performance boosts for applications with random read-intensive workloads.
This document provides an overview of Hyper-V Dynamic Memory in Windows Server 2008 R2 SP1. It discusses how Dynamic Memory allows virtual machines to dynamically adjust their memory allocation based on usage to improve consolidation ratios with minimal performance impact. It covers the architecture including adding and removing memory from VMs, and how Dynamic Memory impacts root memory reserves and NUMA memory management. Alternative memory techniques from VMware like page sharing and second level paging are also briefly discussed.
This document discusses Hyper-V Dynamic Memory, a new feature in Windows Server 2008 R2 SP1 that allows virtual machines to dynamically adjust their memory usage. It aims to improve VM consolidation ratios with minimal performance impact. Dynamic Memory architecture is explained, including how memory is added and removed from VMs. The impact on root memory reserves and NUMA memory management is covered. Alternative memory techniques from VMware are also discussed.
The document discusses solid state drives (SSDs) and their benefits over traditional hard disk drives (HDDs). SSDs offer much faster read/write speeds and lower power consumption compared to HDDs. A mixed configuration using both SSDs and HDDs is typically recommended, with "hot" frequently accessed data stored on SSDs and "cold" infrequently accessed data stored on HDDs. The document provides examples of workloads and applications that are good fits for SSDs, and also discusses technologies used in SSDs like single-level cells and multi-level cells. It also covers how to identify and move hot data to SSDs for the AIX and IBM i operating systems.
This document summarizes a presentation on datacenter computing trends and problems. It discusses how cooling is a major source of energy inefficiency in datacenters. It also explains how servers are rarely fully utilized but operate least efficiently during common usage of 30% load. The document advocates for achieving better energy proportionality so servers can be more efficient during typical usage levels. It presents approaches like disaggregated memory and servers that break CPU-memory co-location to improve efficiency and consolidation.
The Dell Latitude D610 is a durable and flexible notebook designed for mobile professionals. Weighing less than 5 pounds, it delivers powerful performance from an Intel Pentium M processor while withstanding tough conditions through its magnesium alloy frame and sealed keyboard. It provides connectivity and expandability through features like wireless LAN, Gigabit Ethernet, USB ports, and an optional docking station.
This is a paper was written by David Reine, an IT analyst for The Clipper Group, and highlights IBM’s SAN Volume Controller new features, capabilities and benefits. These new capabilities were announced on October 20, 2009 If you have a heterogeneous storage architecture in your data center that is under-utilized and costing the enterprise on the bottom line, IBM SVC 5 may be the solution that you have
The document introduces EMC's new Data Domain DD800 appliance series and Data Domain Archiver for backup and archive storage. The systems provide faster backup speeds, increased capacity, and cost-effective long-term retention compared to traditional tape storage. The appliances leverage data deduplication and support all major backup software for backup, archive, and disaster recovery across on-premise and off-premise locations.
This document defines key concepts related to virtualization and disk performance. It discusses how virtualization can compound disk fragmentation issues and slow performance. The disk subsystem is identified as the main performance bottleneck for virtualized environments due to the additional processing layers from guest to host systems. Best practices for improving disk performance in virtualized servers include advanced defragmentation of host and guest systems, adjusting filesystem settings, separating disks and partitions, and using high-performance storage.
The document describes several common computer hardware components. It discusses hard disks for secondary storage, central processing units (CPUs) that perform computations, random access memory (RAM) for temporary data storage, graphics cards that generate video output, optical disc drives like CD-ROMs, motherboards that connect major components, and more. The document provides details on the purpose and basic functioning of each part.
This document provides an overview of the Distributed Symmetric Multiprocessing (DSMP) software architecture. DSMP transforms an InfiniBand connected cluster of commodity servers into a shared memory supercomputer through two unique software components: 1) a host operating system that runs on the head node, and 2) a lightweight microkernel that runs on the other servers. Key aspects of DSMP include a shared memory system, optimized InfiniBand drivers, an application-driven memory page coherency scheme, enhanced multithreading support, and distributed disk storage. DSMP allows commodity clusters to provide shared memory capabilities at a lower cost than proprietary supercomputers.
Practical experiences and best practices for SSD and IBM iCOMMON Europe
Â
This document discusses using solid state drives (SSDs) to optimize performance on IBM Power Systems. It provides examples showing that SSDs can significantly reduce batch window times, lower response times for transactions, and accelerate analytics. SSDs offer much faster read/write speeds than hard disk drives (HDDs) and can cut batch windows by 40-50%. For queries on data warehouses, SSDs deliver reductions of 60-87% in processing time. SSDs also allow for 85% faster data building and compression for deep analytics. Mixed SSD/HDD configurations with SSDs holding hot data can deliver the best price/performance.
Genie Backup Manager es la solucion para respaldos favorita de los profesionales de TI para el respaldo automatizado de estaciones y servidores en la pequeña y mediana empresa.
This document provides an overview and summary of key concepts around virtualization that will be covered in more depth at a technical deep dive session, including:
- Virtualization capabilities for desktops/laptops and servers including workstation virtualization and server consolidation.
- How virtual machines work and the overhead associated with virtualization.
- Properties of virtualization like partitioning, isolation, and encapsulation.
- Benefits of server virtualization like consolidation, simpler management, and automated resource pooling.
- Comparison of "hosted" and vSphere virtualization architectures.
- Technologies used in virtualization like binary translation, hardware assistance from Intel VT/AMD-V.
- Ability to virtualize CPU intensive applications with
Memory virtualization allows virtual machines to access virtual memory addresses that are translated to physical memory addresses. Hardware support for memory virtualization reduces overhead by offloading page table management to the CPU. Memory sharing between virtual machines reduces memory usage by identifying identical pages and having them share physical memory. Virtual memory overcommitment allocates more virtual memory than physical memory available by swapping out unused memory to disk. Techniques for memory sharing and overcommitment aim to improve memory utilization in virtualized systems.
Version 1 of DPM provided disk-based replication of files and centralized backup of branch offices. Version 2 added seamless disk and tape protection for Windows file and application servers along with system state and cluster support. Version 3 unified backup to disk, tape and cloud with additional protection for advanced Microsoft workloads, Windows clients and enterprise scalability.
The document discusses memory management in operating systems. It describes how operating systems allocate memory to processes and track which memory is in use and which is free. It discusses different memory management techniques including fixed partitions, dynamic partitioning with variable sized partitions, and compaction to reduce fragmentation. It also summarizes different dynamic partitioning placement algorithms like first-fit, next-fit, best-fit, and worst-fit.
Evaluation and Enhancement to Memory Sharing and Swapping in Xen 4.1The Linux Foundation
Â
This document evaluates and enhances memory sharing and swapping features in Xen 4.1. It summarizes the current memory overcommit features in Xen, including blktap2 sharing, 0-page sharing, and host swapping. It proposes enhancements to these features and designs a memory overcommit policy to achieve higher VM density without sacrificing performance. Experimental results show that the combination of blktap2 sharing and 0-page sharing is effective with little performance overhead. A policy that prioritizes swapping to sharing improves performance over default swapping policies. The enhanced features allow increasing VM density in a VDI scenario by over 70% while maintaining quality of experience.
Structure for scale: Dialing in your apps for optimal performanceAtlassian
Â
The document discusses upgrading a JIRA instance to improve performance by upgrading hardware including adding more CPUs, memory, and disk space as well as upgrading software including Java, the operating system, and database. It provides details on the specific hardware and software upgrades made and shows significant performance improvements such as reducing reindex times by 59% and full garbage collections by over 80%.
The Dell OptiPlex 3010 is an affordable business desktop that delivers the latest Intel processors and graphics for office productivity. It comes in mini-tower, desktop, and small form factor chassis to maximize desk space. The durable metal chassis underwent testing for reliability. It can be managed easily with standard solutions and has features for connectivity, security, and productivity.
The document discusses Datacore SANsymphony-V, a software-defined storage solution that provides virtualization, pooling and provisioning, continuous availability, performance acceleration, backup/recovery, and remote replication capabilities. It summarizes key features like virtualization of servers, storage and desktops; infrastructure management functions; benefits like cost savings and high utilization; and integration with VMware vSphere.
The document discusses extending the use of Xen into embedded and communications workloads. It covers various usage models including robotics, IP phones, and routing. Key virtual machine monitor requirements are discussed for different segments including industrial control, communications appliances, and benchmarks. Challenges around scheduling, traffic prioritization, power management, and direct cache access are also summarized.
XPDS16: Patch review for non-maintainers - George Dunlap, Citrix Systems R&D...The Linux Foundation
Â
As the number of contributions grow, reviewer bandwidth becomes a bottleneck; and maintainers are always asking for more help. However,
ultimately maintainers must at least Ack every patch that goes in; so if you're not a maintainer, how can you contribute? Why should anyone care about your opinion?
This talk will try to lay out some advice and guidelines for non-maintainers, for how they can do code review in a way which will effectively reduce the load on maintainers when they do come to review a patch.
Virtualization is becoming increasingly pervasive due to server consolidation and new uses like security and high availability. Virtualization reduces costs by improving server utilization. Future trends include live migration between servers, I/O virtualization using AMD's IOMMU to assign devices directly to virtual machines, and nested virtualization to run hypervisors inside guest virtual machines. AMD's new IOMMUv2 adds support for demand paging of devices using page faults, allowing more efficient use of memory.
Migration of virtual machines without guest downtime is a key feature for hypervisors. Sadly, not all hardware is the same, and keeping guests running in a heterogeneous environment takes a lot of care. Normally, features are advertised via the CPUID instruction, but life is never as simple as we would like. Andrew will discuss what information needs to be controlled, what information can and can't be controlled, and how it applies to Xen guests.
Libxenlight is a C library that provides a simple API for toolstacks like Xend and Libvirt to perform Xen operations. It aims to consolidate the code for different toolstacks, which currently have duplications and inefficiencies. Libxenlight follows design principles like being stateless and hiding lower-level Xen components. A new toolstack called XL has also been created with the goals of being small, tested and compatible with xm commands while using Libxenlight. Multiple toolstacks may be ported to use Libxenlight and its development is open to contributions.
This is a paper was written by David Reine, an IT analyst for The Clipper Group, and highlights IBM’s SAN Volume Controller new features, capabilities and benefits. These new capabilities were announced on October 20, 2009 If you have a heterogeneous storage architecture in your data center that is under-utilized and costing the enterprise on the bottom line, IBM SVC 5 may be the solution that you have
The document introduces EMC's new Data Domain DD800 appliance series and Data Domain Archiver for backup and archive storage. The systems provide faster backup speeds, increased capacity, and cost-effective long-term retention compared to traditional tape storage. The appliances leverage data deduplication and support all major backup software for backup, archive, and disaster recovery across on-premise and off-premise locations.
This document defines key concepts related to virtualization and disk performance. It discusses how virtualization can compound disk fragmentation issues and slow performance. The disk subsystem is identified as the main performance bottleneck for virtualized environments due to the additional processing layers from guest to host systems. Best practices for improving disk performance in virtualized servers include advanced defragmentation of host and guest systems, adjusting filesystem settings, separating disks and partitions, and using high-performance storage.
The document describes several common computer hardware components. It discusses hard disks for secondary storage, central processing units (CPUs) that perform computations, random access memory (RAM) for temporary data storage, graphics cards that generate video output, optical disc drives like CD-ROMs, motherboards that connect major components, and more. The document provides details on the purpose and basic functioning of each part.
This document provides an overview of the Distributed Symmetric Multiprocessing (DSMP) software architecture. DSMP transforms an InfiniBand connected cluster of commodity servers into a shared memory supercomputer through two unique software components: 1) a host operating system that runs on the head node, and 2) a lightweight microkernel that runs on the other servers. Key aspects of DSMP include a shared memory system, optimized InfiniBand drivers, an application-driven memory page coherency scheme, enhanced multithreading support, and distributed disk storage. DSMP allows commodity clusters to provide shared memory capabilities at a lower cost than proprietary supercomputers.
Practical experiences and best practices for SSD and IBM iCOMMON Europe
Â
This document discusses using solid state drives (SSDs) to optimize performance on IBM Power Systems. It provides examples showing that SSDs can significantly reduce batch window times, lower response times for transactions, and accelerate analytics. SSDs offer much faster read/write speeds than hard disk drives (HDDs) and can cut batch windows by 40-50%. For queries on data warehouses, SSDs deliver reductions of 60-87% in processing time. SSDs also allow for 85% faster data building and compression for deep analytics. Mixed SSD/HDD configurations with SSDs holding hot data can deliver the best price/performance.
Genie Backup Manager es la solucion para respaldos favorita de los profesionales de TI para el respaldo automatizado de estaciones y servidores en la pequeña y mediana empresa.
This document provides an overview and summary of key concepts around virtualization that will be covered in more depth at a technical deep dive session, including:
- Virtualization capabilities for desktops/laptops and servers including workstation virtualization and server consolidation.
- How virtual machines work and the overhead associated with virtualization.
- Properties of virtualization like partitioning, isolation, and encapsulation.
- Benefits of server virtualization like consolidation, simpler management, and automated resource pooling.
- Comparison of "hosted" and vSphere virtualization architectures.
- Technologies used in virtualization like binary translation, hardware assistance from Intel VT/AMD-V.
- Ability to virtualize CPU intensive applications with
Memory virtualization allows virtual machines to access virtual memory addresses that are translated to physical memory addresses. Hardware support for memory virtualization reduces overhead by offloading page table management to the CPU. Memory sharing between virtual machines reduces memory usage by identifying identical pages and having them share physical memory. Virtual memory overcommitment allocates more virtual memory than physical memory available by swapping out unused memory to disk. Techniques for memory sharing and overcommitment aim to improve memory utilization in virtualized systems.
Version 1 of DPM provided disk-based replication of files and centralized backup of branch offices. Version 2 added seamless disk and tape protection for Windows file and application servers along with system state and cluster support. Version 3 unified backup to disk, tape and cloud with additional protection for advanced Microsoft workloads, Windows clients and enterprise scalability.
The document discusses memory management in operating systems. It describes how operating systems allocate memory to processes and track which memory is in use and which is free. It discusses different memory management techniques including fixed partitions, dynamic partitioning with variable sized partitions, and compaction to reduce fragmentation. It also summarizes different dynamic partitioning placement algorithms like first-fit, next-fit, best-fit, and worst-fit.
Evaluation and Enhancement to Memory Sharing and Swapping in Xen 4.1The Linux Foundation
Â
This document evaluates and enhances memory sharing and swapping features in Xen 4.1. It summarizes the current memory overcommit features in Xen, including blktap2 sharing, 0-page sharing, and host swapping. It proposes enhancements to these features and designs a memory overcommit policy to achieve higher VM density without sacrificing performance. Experimental results show that the combination of blktap2 sharing and 0-page sharing is effective with little performance overhead. A policy that prioritizes swapping to sharing improves performance over default swapping policies. The enhanced features allow increasing VM density in a VDI scenario by over 70% while maintaining quality of experience.
Structure for scale: Dialing in your apps for optimal performanceAtlassian
Â
The document discusses upgrading a JIRA instance to improve performance by upgrading hardware including adding more CPUs, memory, and disk space as well as upgrading software including Java, the operating system, and database. It provides details on the specific hardware and software upgrades made and shows significant performance improvements such as reducing reindex times by 59% and full garbage collections by over 80%.
The Dell OptiPlex 3010 is an affordable business desktop that delivers the latest Intel processors and graphics for office productivity. It comes in mini-tower, desktop, and small form factor chassis to maximize desk space. The durable metal chassis underwent testing for reliability. It can be managed easily with standard solutions and has features for connectivity, security, and productivity.
The document discusses Datacore SANsymphony-V, a software-defined storage solution that provides virtualization, pooling and provisioning, continuous availability, performance acceleration, backup/recovery, and remote replication capabilities. It summarizes key features like virtualization of servers, storage and desktops; infrastructure management functions; benefits like cost savings and high utilization; and integration with VMware vSphere.
The document discusses extending the use of Xen into embedded and communications workloads. It covers various usage models including robotics, IP phones, and routing. Key virtual machine monitor requirements are discussed for different segments including industrial control, communications appliances, and benchmarks. Challenges around scheduling, traffic prioritization, power management, and direct cache access are also summarized.
XPDS16: Patch review for non-maintainers - George Dunlap, Citrix Systems R&D...The Linux Foundation
Â
As the number of contributions grow, reviewer bandwidth becomes a bottleneck; and maintainers are always asking for more help. However,
ultimately maintainers must at least Ack every patch that goes in; so if you're not a maintainer, how can you contribute? Why should anyone care about your opinion?
This talk will try to lay out some advice and guidelines for non-maintainers, for how they can do code review in a way which will effectively reduce the load on maintainers when they do come to review a patch.
Virtualization is becoming increasingly pervasive due to server consolidation and new uses like security and high availability. Virtualization reduces costs by improving server utilization. Future trends include live migration between servers, I/O virtualization using AMD's IOMMU to assign devices directly to virtual machines, and nested virtualization to run hypervisors inside guest virtual machines. AMD's new IOMMUv2 adds support for demand paging of devices using page faults, allowing more efficient use of memory.
Migration of virtual machines without guest downtime is a key feature for hypervisors. Sadly, not all hardware is the same, and keeping guests running in a heterogeneous environment takes a lot of care. Normally, features are advertised via the CPUID instruction, but life is never as simple as we would like. Andrew will discuss what information needs to be controlled, what information can and can't be controlled, and how it applies to Xen guests.
Libxenlight is a C library that provides a simple API for toolstacks like Xend and Libvirt to perform Xen operations. It aims to consolidate the code for different toolstacks, which currently have duplications and inefficiencies. Libxenlight follows design principles like being stateless and hiding lower-level Xen components. A new toolstack called XL has also been created with the goals of being small, tested and compatible with xm commands while using Libxenlight. Multiple toolstacks may be ported to use Libxenlight and its development is open to contributions.
The ARM architecture strongly recommends to use a break-before-make when changing translation table entries whenever certain conditions are met. Failing to do so may result in getting TLB conflicts or breaking the coherency.
During this session, we will introduce break-before-make and when the code handling page tables should use it. We will also discuss the modifications required in Xen to avoid breaking the coherency.
Adding support for you new shiny board in Xen on ARM is a simple task once you get a kernel running on bare metal.
This session will cover the different steps to port Xen on ARM from the firmware to the shell prompt in DOM0.
We will give you tips on the common pitfalls when you have your hypervisor, or your DOM0 kernel crashing. We will also provide suggestion on how to debug when the console is not working.
XPDS16: High-Performance Virtualization for HPC Cloud on Xen - Jun Nakajima &...The Linux Foundation
Â
We have been working to get Xen up and running on self-boot Intel® Xeon Phi processors to build HPC clouds. We see several challenges because of the unique (but not unusual for HPC) hardware technologies and performance requirements. For example, such hardware technologies include 1) >256 CPUs, 2) MCDRAM (high-bandwidth memory), 3) integrated fabric (i.e. Intel® Omni-Path). Unlike the “coprocessor“ model, supporting self-boot with >256 CPUs has various implications to Xen, including scheduling and scalability. We need to allow user applications to use MCDRAM directly to perform optimally. Also, we need to enable the integrated HPC fabric for the VM to use by direct I/O assignment.
In addition, we have only a single VM on each node to meet the high-performance requirements of HPC clouds. This (i.e. non-shared) model allowed us to optimize Xen more. In this talk, we share our design and lessons, and discuss the options we considered to achieve high-performance virtualization for HPC.
This document summarizes a presentation on memory overcommitment in virtualization given by Dan Magenheimer at the 2008 Xen Summit. It discusses why Xen currently does not support memory overcommitment while other virtualization platforms like VMware do. It then explores possible techniques for implementing memory overcommitment in Xen, such as ballooning, page sharing, and demand paging. The goal would be to allow more efficient memory utilization and higher server consolidation ratios.
XenServer 5.5 - Czy można zaoszczędzić na wirtualizacji serwerów? Darmowy Xen...Peter Ocasek
Â
The document discusses server virtualization using XenServer. It summarizes XenServer features like the Xen hypervisor, support for Intel and AMD processors, storage integration using StorageLink, high availability, workload balancing, and lab/stage management. It also describes the Citrix Essentials product which provides additional features like snapshots, backup support, and Active Directory integration for XenServer.
The document proposes optimizations for crash dump in virtualized environments including performing core dump and system recovery concurrently to reduce downtime, selectively dumping only non-empty memory pages of a crashed virtual machine to improve dump speed, and controlling disk I/O rates between concurrent dump and recovery processes to enhance quality of service.
OpenStorage with OpenStack allows for storage software to be open source without vendor lock-in. NexentaStor is a leading OpenStorage solution that provides file and block access running on standard hardware. It utilizes technologies like deduplication, compression, thin provisioning, and triple parity RAID to provide storage efficiency and data protection. NexentaStor integrates with OpenStack Nova to provision and manage volumes, leveraging cloning to quickly create new volumes from snapshots. Cloud storage use cases like disaster recovery and long term archives can benefit from using NexentaStor with OpenStack.
OpenStorage with OpenStack allows storage software to be open source without vendor lock-in. It uses open data formats and some or all of the components no longer need to come from the same vendor. NexentaStor is a leading OpenStorage solution that can be used with OpenStack Nova to provision volumes from snapshots for virtual machines efficiently using cloning. NexentaStor provides features like deduplication, compression, thin provisioning and triple parity RAID that are beneficial for cloud storage and virtual environments.
This document discusses memory management in operating systems. It defines memory management as allocating RAM to user programs and reclaiming memory after programs finish. It also describes protecting each user's memory from other programs. The document discusses physical and virtual memory, and types of virtual memory including paged, segmented, and swapped memory. It defines static and dynamic memory allocation, with static allocation assigning fixed memory at compile time and dynamic allocation assigning variable memory as needed from the heap.
IBM's 5th generation eX5 server architecture delivers innovations that address key customer pain points around memory limitations and licensing costs. The eX5 introduces MAX5 memory expansion technology that allows servers to double their memory capacity without performance impacts. This enables customers to consolidate more workloads on fewer servers to reduce licensing fees. The eX5 also uses SSD technology through the FlashPack and NEXT IO offerings to provide thousands of IOPS of performance previously requiring hundreds of spinning disks.
A Paradigm Shift: The Increasing Dominance of Memory-Oriented Solutions for H...Ben Stopford
Â
This document discusses the increasing dominance of memory-oriented solutions for high-performance data access. It notes that database lookups are around 20 milliseconds while hashmap lookups are around 20 nanoseconds. It then discusses how abstraction improves software but hurts performance. It outlines the traditional database architecture with disk storage and compares it to newer in-memory and distributed in-memory architectures that can provide faster performance by avoiding disk I/O and leveraging memory and distribution.
[Harvard CS264] 05 - Advanced-level CUDA Programmingnpinto
Â
The document discusses optimizations for memory and communication in massively parallel computing. It recommends caching data in faster shared memory to reduce loads and stores to global device memory. This can improve performance by avoiding non-coalesced global memory accesses. The document provides an example of coalescing writes for a matrix transpose by first loading data into shared memory and then writing columns of the tile to global memory in contiguous addresses.
Similar to Transcendent memoryupdate xensummit2010-final (11)
Static partitioning is used to split an embedded system into multiple domains, each of them having access only to a portion of the hardware on the SoC. It is key to enable mixed-criticality scenarios, where a critical application, often based on a small RTOS, runs alongside a larger non-critical app, typically based on Linux. The two domains cannot interfere with each other.
This talk will explain how to use Xen for static partitioning. It will introduce dom0-less, a new Xen feature written for the purpose. Dom0-less allows multiple VMs to start at boot time directly from the Xen hypervisor, decreasing boot times drastically. It makes it very easy to partition the system without virtualization overhead. Dom0 becomes unnecessary.
This presentation will go into details on how to setup a Xen dom0-less system. It will show configuration examples and explain device assignment. The talk will discuss its implications for latency-sensitive and safety-critical environments.
XPDDS19: How TrenchBoot is Enabling Measured Launch for Open-Source Platform ...The Linux Foundation
Â
Daniel Smith discusses TrenchBoot, a project aiming to establish a unified approach to harnessing boot integrity technologies across open source platforms. TrenchBoot will enable establishing hardware-rooted integrity during platform boot (first launch inspection), runtime (runtime inspection), and other states. For runtime inspection, TrenchBoot will develop a way to securely re-establish the integrity of Xen at any time without rebooting by dynamically launching an integrity kernel to inspect and verify Xen. The talk outlines the initial and future work of TrenchBoot to integrate these capabilities with Linux, Xen and other open source projects.
XPDDS19 Keynote: Xen in Automotive - Artem Mygaiev, Director, Technology Solu...The Linux Foundation
Â
Artem will briefly cover what has been done since the first talk on Xen in Automotive domain back in 2013, what is going on now and what is still missing for broad adaptation of Xen in vehicles. The following topics will be covered:
Embedded/automotive features of Xen
Collaboration with AGL and GENIVI organizations for standardization
Efforts on Functional Safety compliance
Artem will also go over typical automotive use scenarios for Xen which may not be the same as generic computing use of hypervisor.
XPDDS19 Keynote: Xen Project Weather Report 2019 - Lars Kurth, Director of Op...The Linux Foundation
Â
In this keynote talk, we will give an overview of the state of the Xen Project, trends that impact the project, see whether challenges that surfaced last year have been addressed and how we did it, and highlight new challenges and solutions for the coming year.
In​ ​recent​ ​years unikernels have​ ​shown​ ​immense​ ​performance potential​ (e.g., boot times of only a few ms, image sizes of only hundreds of KBs).The​ ​fundamental​ ​drawback​ ​of​ ​unikernels​ ​is​ ​that​ ​they​ ​require​ ​that applications​ ​be​ ​manually​ ​ported​ ​to​ ​the​ ​underlying​ ​minimalistic​ ​OS, needing​ ​both​ ​expert​ ​work​ ​and​ ​often​ ​considerable​ ​amount of​ ​time.​
The Unikraft project provides a unikernel code base and build system that significantly simplifies the building of unikernels. In addition to support for a number CPU architectures, languages and frameworks, Unikraft provides debugging and tracing features that are generally sorely missing from unikernel projects. In this talk we will talk about these features, show a set of preliminary performance numbers, and provide a roadmap for the project's future.
XPDDS19 Keynote: Secret-free Hypervisor: Now and Future - Wei Liu, Software E...The Linux Foundation
Â
The idea of making Xen secret-free has been floating since Spectre and Meltdown came into light. In this talk we will discuss what is being done and what needs to be done next.
XPDDS19 Keynote: Xen Dom0-less - Stefano Stabellini, Principal Engineer, XilinxThe Linux Foundation
Â
This talk will introduce Dom0-less: a new way of using Xen to build mixed-criticality solutions. Dom0-less is a Xen feature that adds a novel approach to static partitioning based on virtualization. It allows multiple domains to start at boot time directly from the Xen hypervisor, decreasing boot times dramatically. Xen userspace tools, such as xl and libvirt, become optional.
Dom0-less extends the existing device tree based Xen boot protocol to cover information required by additional domains. Binaries, such as kernels and ramdisks, are loaded by the bootloader (u-boot) and advertised to Xen via new device tree bindings.
The audience will learn how to use Dom0-less to partition the system. Uboot and device tree configuration details will be explained to enable the audience to get the most out of this feature. The talk will include a status update and details on future plans.
XPDDS19 Keynote: Patch Review for Non-maintainers - George Dunlap, Citrix Sys...The Linux Foundation
Â
As the number of contributions grow, reviewer bandwidth becomes a bottleneck; and maintainers are always asking for more help. However, ultimately maintainers must at least Ack every patch that goes in; so if you're not a maintainer, how can you contribute? Why should anyone care about your opinion?
This talk will try to lay out some advice and guidelines for non-maintainers, for how they can do code review in a way which will effectively reduce the load on maintainers when they do come to review a patch.
This talk is a follow-up to our Summit 2017 presentation in which we covered our plans for Intel VMFUNC and #VE, as well as related use-cases. This year, we will provide a report on what we have accomplished in Xen 4.12, and what remains to be addressed. We will also give a brief status update of VMI on AMD hardware. The session will end with some real-world numbers of the Hypervisor Introspection solution running on Citrix Hypervisor 8.0 with #VE enabled.
OSSJP/ALS19: The Road to Safety Certification: Overcoming Community Challeng...The Linux Foundation
Â
Safety certification is one of the essential requirements for software to be used in highly regulated industries. Besides technical and compliance issues (such as ISO 26262 vs IEC 611508) transitioning an existing project to become more easily safety certifiable requires significant changes to development practices within an open source project.
In this session, we will lay out some challenges of making safety certification achievable in open source and the Xen Project. We will outline the process the Xen Project has followed thus far and highlight lessons learned along the way. The talk will primarily focus on necessary process, tooling changes and community challenges that can prevent progress. We will be offering an in-depth review of how Xen Project is approaching this challenging goal and try to derive lessons for other projects and contributors.
OSSJP/ALS19: The Road to Safety Certification: How the Xen Project is Making...The Linux Foundation
Â
This document summarizes a discussion around enabling functional safety certification for the Xen open source hypervisor project. Key points discussed include:
- Establishing a split development model with open and closed parts to balance community needs and safety requirements.
- Developing reference implementations and stacks supported by multiple vendors to demonstrate safety certification feasibility.
- Creating plans and processes around requirements, documentation, verification testing, and tooling integration to begin filling gaps for certification.
- Addressing challenges around funding, resources, expertise, and maintaining contributions to ensure any initial work is sustainable long-term.
- Taking an iterative, agile approach to make early progress while further securing necessary funding and support from interested parties.
XPDDS19: Speculative Sidechannels and Mitigations - Andrew Cooper, CitrixThe Linux Foundation
Â
2018 saw fundamental shifts in security boundaries which were previously taken for granted. A lot of work has been done in the past 2 years, and largely in secret under embargo, but there is plenty more work to be done to strengthen the existing mitigations and to try to recover some performance without reopening security holes.
This talk will look at speculative execution sidechannels, the work which has already been done to mitigate the security holes, and future work which hopes to bring some improvements.
XPDDS19: Keeping Coherency on Arm: Reborn - Julien Grall, Arm ltdThe Linux Foundation
Â
The Arm architecture provides a set of guidelines that any software should abide by when accessing the memory with MMU off and update page-tables. Failing to do so may result in getting TLB conflicts or breaking coherency.
In a previous talk ("Keeping coherency on Arm"), we focused on updating safely the stage-2 (aka P2M) page-tables. This talk will focus on the boot code and Xen memory management.
During this session, we will introduce some of the guidelines and when they should be used. We will also discuss how Xen boot sequence needs to be reworked to avoid breaking the guidelines.
XPDDS19: QEMU PV Backend 'qdevification'... What Does it Mean? - Paul Durrant...The Linux Foundation
Â
For many years the QEMU codebase has contained PV backends for Xen guests, giving them paravirtual access to storage, network, keyboard, mouse, etc. however these backends have not been configurable as QEMU devices as their implementation did not fully adhere to the QEMU Object Model (QOM).
Particularly the PV storage backend not using proper QOM devices, or qdevs, meant that the QEMU block layer needed to maintain legacy code that was cluttering up the source. This was causing push-back from the maintainers who did not want to accept any patches relating to that Xen backend until it was 'qdevified'.
In this talk, I'll explain the modifications I made to QEMU to achieve 'qdevification' of the PV storage backend, how compatibility with the libxl toolstack was maintained, and what the next steps in both QEMU and libxl development should be.
PCI is a local computer bus for attaching hardware devices in a computer, and is the main peripheral bus on modern x86 systems. As such, having a proper way to emulate it is crucial for Xen to be able to expose both fully emulated devices or passthrough devices to guests.
This talk will focus on the current status of PCI emulation in Xen, how and where it is used, what are its main limitations and future plans to improve it in order to be more robust and modular.
XPDDS19: [ARM] OP-TEE Mediator in Xen - Volodymyr Babchuk, EPAM SystemsThe Linux Foundation
Â
Volodymyr will speak about TEE mediators. This is a new feature in Xen which allows multiple virtual machines to interact with Trusted Execution Environment available on platform. He developed mediator for one of TEEs, namely OP-TEE.
He will give background information on why TEE is needed at all and share some implementation details.
XPDDS19: Bringing Xen to the Masses: The Story of Building a Community-driven...The Linux Foundation
Â
Xen is a very powerful hypervisor with a talented and diverse developers community. Despite the fact it's almost everywhere (from the Cloud to the embedded world), it can be difficult to set up and manage as a system administrator. General purpose distros have Xen packages, but that's just a start in your Xen journey: you need some tooling and knowledge to have a working and scalable platform.
XCP-ng was built to overcome those issues: by bringing Xen to the masses with a fully turnkey distro with Xen as its core. It's the logical sequel to the XCP project, with a community focus from the start. We'll see how it happened, what we did, and what's next. Finally, we'll see the impact of XCP-ng on the Xen Project.
XPDDS19: Will Robots Automate Your Job Away? Streamlining Xen Project Contrib...The Linux Foundation
Â
Doug has long advocated for more CI/CD (Continuous Integration / Continuous Delivery) processes to be adopted by the Xen Project from the use of Travis CI and now GitLab CI. This talk aims to propose ideas for building upon the existing process and transforming the development process to provide users a higher quality with each release by the Xen Project.
XPDDS19: Client Virtualization Toolstack in Go - Nick Rosbrook & Brendan Kerr...The Linux Foundation
Â
High level toolstacks for server and cloud virtualization are very mature with large communities using and supporting them. Client virtualization is a much more niche community with unique requirements when compared to those found in the server space. In this talk, we’ll introduce a client virtualization toolstack for Xen (redctl) that we are using in Redfield, a new open-source client virtualization distribution that builds upon the work done by the greater virtualization and Linux communities. We will present a case for maturing libxl’s Go bindings and discuss what advantages Go has to offer for high level toolstacks, including in the server space.
Today Xen is scheduling guest virtual cpus on all available physical cpus independently from each other. Recent security issues on modern processors (e.g. L1TF) require to turn off hyperthreading for best security in order to avoid leaking information from one hyperthread to the other. One way to avoid having to turn off hyperthreading is to only ever schedule virtual cpus of the same guest on one physical core at the same time. This is called core scheduling.
This presentation shows results from the effort to implement core scheduling in the Xen hypervisor. The basic modifications in Xen are presented and performance numbers with core scheduling active are shown.
Have you ever been confused by the myriad of choices offered by AWS for hosting a website or an API?
Lambda, Elastic Beanstalk, Lightsail, Amplify, S3 (and more!) can each host websites + APIs. But which one should we choose?
Which one is cheapest? Which one is fastest? Which one will scale to meet our needs?
Join me in this session as we dive into each AWS hosting service to determine which one is best for your scenario and explain why!
Northern Engraving | Nameplate Manufacturing Process - 2024Northern Engraving
Â
Manufacturing custom quality metal nameplates and badges involves several standard operations. Processes include sheet prep, lithography, screening, coating, punch press and inspection. All decoration is completed in the flat sheet with adhesive and tooling operations following. The possibilities for creating unique durable nameplates are endless. How will you create your brand identity? We can help!
Conversational agents, or chatbots, are increasingly used to access all sorts of services using natural language. While open-domain chatbots - like ChatGPT - can converse on any topic, task-oriented chatbots - the focus of this paper - are designed for specific tasks, like booking a flight, obtaining customer support, or setting an appointment. Like any other software, task-oriented chatbots need to be properly tested, usually by defining and executing test scenarios (i.e., sequences of user-chatbot interactions). However, there is currently a lack of methods to quantify the completeness and strength of such test scenarios, which can lead to low-quality tests, and hence to buggy chatbots.
To fill this gap, we propose adapting mutation testing (MuT) for task-oriented chatbots. To this end, we introduce a set of mutation operators that emulate faults in chatbot designs, an architecture that enables MuT on chatbots built using heterogeneous technologies, and a practical realisation as an Eclipse plugin. Moreover, we evaluate the applicability, effectiveness and efficiency of our approach on open-source chatbots, with promising results.
zkStudyClub - LatticeFold: A Lattice-based Folding Scheme and its Application...Alex Pruden
Â
Folding is a recent technique for building efficient recursive SNARKs. Several elegant folding protocols have been proposed, such as Nova, Supernova, Hypernova, Protostar, and others. However, all of them rely on an additively homomorphic commitment scheme based on discrete log, and are therefore not post-quantum secure. In this work we present LatticeFold, the first lattice-based folding protocol based on the Module SIS problem. This folding protocol naturally leads to an efficient recursive lattice-based SNARK and an efficient PCD scheme. LatticeFold supports folding low-degree relations, such as R1CS, as well as high-degree relations, such as CCS. The key challenge is to construct a secure folding protocol that works with the Ajtai commitment scheme. The difficulty, is ensuring that extracted witnesses are low norm through many rounds of folding. We present a novel technique using the sumcheck protocol to ensure that extracted witnesses are always low norm no matter how many rounds of folding are used. Our evaluation of the final proof system suggests that it is as performant as Hypernova, while providing post-quantum security.
Paper Link: https://eprint.iacr.org/2024/257
Introduction of Cybersecurity with OSS at Code Europe 2024Hiroshi SHIBATA
Â
I develop the Ruby programming language, RubyGems, and Bundler, which are package managers for Ruby. Today, I will introduce how to enhance the security of your application using open-source software (OSS) examples from Ruby and RubyGems.
The first topic is CVE (Common Vulnerabilities and Exposures). I have published CVEs many times. But what exactly is a CVE? I'll provide a basic understanding of CVEs and explain how to detect and handle vulnerabilities in OSS.
Next, let's discuss package managers. Package managers play a critical role in the OSS ecosystem. I'll explain how to manage library dependencies in your application.
I'll share insights into how the Ruby and RubyGems core team works to keep our ecosystem safe. By the end of this talk, you'll have a better understanding of how to safeguard your code.
Main news related to the CCS TSI 2023 (2023/1695)Jakub Marek
Â
An English 🇬🇧 translation of a presentation to the speech I gave about the main changes brought by CCS TSI 2023 at the biggest Czech conference on Communications and signalling systems on Railways, which was held in Clarion Hotel Olomouc from 7th to 9th November 2023 (konferenceszt.cz). Attended by around 500 participants and 200 on-line followers.
The original Czech 🇨🇿 version of the presentation can be found here: https://www.slideshare.net/slideshow/hlavni-novinky-souvisejici-s-ccs-tsi-2023-2023-1695/269688092 .
The videorecording (in Czech) from the presentation is available here: https://youtu.be/WzjJWm4IyPk?si=SImb06tuXGb30BEH .
Discover top-tier mobile app development services, offering innovative solutions for iOS and Android. Enhance your business with custom, user-friendly mobile applications.
How information systems are built or acquired puts information, which is what they should be about, in a secondary place. Our language adapted accordingly, and we no longer talk about information systems but applications. Applications evolved in a way to break data into diverse fragments, tightly coupled with applications and expensive to integrate. The result is technical debt, which is re-paid by taking even bigger "loans", resulting in an ever-increasing technical debt. Software engineering and procurement practices work in sync with market forces to maintain this trend. This talk demonstrates how natural this situation is. The question is: can something be done to reverse the trend?
Driving Business Innovation: Latest Generative AI Advancements & Success StorySafe Software
Â
Are you ready to revolutionize how you handle data? Join us for a webinar where we’ll bring you up to speed with the latest advancements in Generative AI technology and discover how leveraging FME with tools from giants like Google Gemini, Amazon, and Microsoft OpenAI can supercharge your workflow efficiency.
During the hour, we’ll take you through:
Guest Speaker Segment with Hannah Barrington: Dive into the world of dynamic real estate marketing with Hannah, the Marketing Manager at Workspace Group. Hear firsthand how their team generates engaging descriptions for thousands of office units by integrating diverse data sources—from PDF floorplans to web pages—using FME transformers, like OpenAIVisionConnector and AnthropicVisionConnector. This use case will show you how GenAI can streamline content creation for marketing across the board.
Ollama Use Case: Learn how Scenario Specialist Dmitri Bagh has utilized Ollama within FME to input data, create custom models, and enhance security protocols. This segment will include demos to illustrate the full capabilities of FME in AI-driven processes.
Custom AI Models: Discover how to leverage FME to build personalized AI models using your data. Whether it’s populating a model with local data for added security or integrating public AI tools, find out how FME facilitates a versatile and secure approach to AI.
We’ll wrap up with a live Q&A session where you can engage with our experts on your specific use cases, and learn more about optimizing your data workflows with AI.
This webinar is ideal for professionals seeking to harness the power of AI within their data management systems while ensuring high levels of customization and security. Whether you're a novice or an expert, gain actionable insights and strategies to elevate your data processes. Join us to see how FME and AI can revolutionize how you work with data!
Digital Banking in the Cloud: How Citizens Bank Unlocked Their MainframePrecisely
Â
Inconsistent user experience and siloed data, high costs, and changing customer expectations – Citizens Bank was experiencing these challenges while it was attempting to deliver a superior digital banking experience for its clients. Its core banking applications run on the mainframe and Citizens was using legacy utilities to get the critical mainframe data to feed customer-facing channels, like call centers, web, and mobile. Ultimately, this led to higher operating costs (MIPS), delayed response times, and longer time to market.
Ever-changing customer expectations demand more modern digital experiences, and the bank needed to find a solution that could provide real-time data to its customer channels with low latency and operating costs. Join this session to learn how Citizens is leveraging Precisely to replicate mainframe data to its customer channels and deliver on their “modern digital bank” experiences.
[OReilly Superstream] Occupy the Space: A grassroots guide to engineering (an...Jason Yip
Â
The typical problem in product engineering is not bad strategy, so much as “no strategy”. This leads to confusion, lack of motivation, and incoherent action. The next time you look for a strategy and find an empty space, instead of waiting for it to be filled, I will show you how to fill it in yourself. If you’re wrong, it forces a correction. If you’re right, it helps create focus. I’ll share how I’ve approached this in the past, both what works and lessons for what didn’t work so well.
Generating privacy-protected synthetic data using Secludy and MilvusZilliz
Â
During this demo, the founders of Secludy will demonstrate how their system utilizes Milvus to store and manipulate embeddings for generating privacy-protected synthetic data. Their approach not only maintains the confidentiality of the original data but also enhances the utility and scalability of LLMs under privacy constraints. Attendees, including machine learning engineers, data scientists, and data managers, will witness first-hand how Secludy's integration with Milvus empowers organizations to harness the power of LLMs securely and efficiently.
Ivanti’s Patch Tuesday breakdown goes beyond patching your applications and brings you the intelligence and guidance needed to prioritize where to focus your attention first. Catch early analysis on our Ivanti blog, then join industry expert Chris Goettl for the Patch Tuesday Webinar Event. There we’ll do a deep dive into each of the bulletins and give guidance on the risks associated with the newly-identified vulnerabilities.
1. <Insert Picture Here>
Update on
Transcendent
2010 Memory on Xen
Speaker: Dan Magenheimer
Oracle Corporation
2. Agenda
• Motivation and Challenge
• BRIEF overview of Physical Memory Management
• BRIEF Transcendent Memory (“tmem”) Overview
• Tmem Progress since Xen Summit 2009
• Self-ballooning + Tmem Performance Analysis
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
3. Motivation
• Memory is increasingly becoming a
bottleneck in virtualized system
• Existing mechanisms have major holes
Four underutilized 2-cpu virtual servers
ballooning
each with 1GB RAM
One 4-CPU physical
server w/4GB RAM
X
X
memory
overcommitment page
sharing
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
4. The Virtualized Physical Memory
Resource Optimization Challenge
Optimize, across time, the distribution of machine
memory among a maximal set of virtual machines by:
• measuring the current and future memory need of
each running VM and
• reclaiming memory from those VMs that have an
excess of memory and either:
• providing it to VMs that need more memory or
• using it to provision additional new VMs.
• without suffering a significant performance penalty
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
5. Agenda
• Motivation and Challenge
• BRIEF Overview of Physical Memory Management
• in an operating system
• in a virtual machine monitor (Xen)
• BRIEF Transcendent Memory (“tmem”) Overview
• Tmem Progress since Xen Summit 2009
• Self-ballooning + Tmem Performance Analysis
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
6. OS Physical Memory Management
• Operating systems
are memory hogs!
OS
Memory constraint
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
7. OS Physical Memory Management
• Operating systems are
memory hogs!
OS
If you give an
operating system
more memory…..
New larger memory
constraint
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
8. OS Physical Memory Management
• Operating systems are
memory hogs!
My name is
Linux and I If you give an OS more
am a
memory hog memory
…it uses up any
memory you give it!
Memory constraint
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
9. OS Physical Memory Management
• What does an OS do
with all that memory?
• Kernel code and data
Kernel code • User code and data
User data
Page • Page cache!
Kernel
User code cache
data
Everything
else
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
10. OS Physical Memory Management
• What does an OS do
with all that memory?
page
cache
Page cache attempts to
predict future needs of
pages from the disk…
sometimes it gets it right
“good” pages
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
11. OS Physical Memory Management
• What does an OS do
with all that memory?
page
cache
Page cache attempts to
predict future needs of
pages from the disk…
sometimes it gets it wrong
“wasted” pages
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
12. OS Physical Memory Management
• What does an OS do
with all that memory?
page cache
…much of the time
mostly page cache
… some of which will
be useful in the future
… and some (or
maybe most…)
of which is wasted
Everything else
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
13. Agenda
• Motivation and Challenge
• BRIEF Overview of Physical Memory Management
• in an operating system
• in a virtual machine monitor (Xen)
• BRIEF Transcendent Memory (“tmem”) Overview
• Tmem Progress since Xen Summit 2009
• Self-ballooning + Tmem Performance Analysis
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
14. VMM Physical Memory Management
fallow
• Xen partitions memory
among more guests
gues fallow
guest t • Xen memory
• dom0 memory
guest
• guest 1 memory
• guest 2 memory
• guest 3…
fallow
guest
• BUT still fallow memory
leftover
fallow, adj., land left without a
crop for one or more years
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
15. VMM Physical Memory Management
in the presence of migration
fallow
gue fallow
guest t
s
Physical gues
machine “B”
• migration t
• requires fallow memory fallow
in the target machine
• leaves behind fallow fallow
memory in the gue
st
fallow
originating machine gues
t
fallow
Physical
guest
machine “A”
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
16. VMM Physical Memory Management
in the presence of ballooning
guest
fallow
Ballooning works great for
guest
giving more memory TO
a guest OS…
guest
Look ma! No more
gues fallow memory! (*burp*)
t
gues
t
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
17. VMM Physical Memory Management
in the presence of ballooning
But not so great to take memory away
• not instantaneous (memory inertia)
• guest can’t predict future needs
• good pages are evicted along with the bad
• host doesn’t know how much/fast to balloon
• too much or too fast has dire results
thrashing or the dreaded OOM killer
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
18. VMM Physical Memory Management
in the presence of migration AND ballooning
• Look ma! No more
fallow
guest fallow memory!
guest
….But now live migration
guest
crashes and burns
gue
st gue
st
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
19. Why this IS a hard problem!
Summary
• OS’s use as much memory as they are given
• but cannot predict the future so often guess wrong
• and often much memory owned by an OS is wasted
• Xen leaves large amounts of memory fallow
• fixed partitioning results in fragmentation
• migration requires fallow memory to succeed
• Ballooning helps but:
• can’t predict future memory needs of guests
• memory has inertia
• the price of incorrect guesses can be dire
NEED A NEW APPROACH TO VIRTUALIZED
PHYSICAL MEMORY MANAGEMENT!!
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
20. Agenda
• Motivation and Challenge
• BRIEF Overview of Physical Memory Management
• BRIEF Transcendent Memory (“tmem”) Overview
• Tmem Progress since Xen Summit 2009
• Self-ballooning + Tmem Performance Analysis
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
21. Transcendent memory
creating the transcendent memory pool
fallow
• Step 1a: reclaim all fallow memory
• Step 1b: reclaim wasted guest
fallow
t
guest gues
memory (e.g. via self-ballooning)
• Step 1c: collect it all into a pool
guest
fallow
guest
Transcendent
memory
pool
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
22. Transcendent memory
creating the transcendent memory pool
• Step 2: provide indirect
guest access, strictly controlled by
t
gues the hypervisor and dom0
data
data
control
Transcendent guest
memory data
pool
data control
guest
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
23. Transcendent memory
API characteristics
Transcendent memory API
guest guest
• paravirtualized (lightly)
• narrow
• well-specified
• operations are:
• synchronous
• page-oriented (one page per op)
• copy-based
• multi-faceted
Transcendent • extensible
memory
pool
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
24. Transcendent memory
four different subpool types
four different uses
Legend:
flags ephemeral persistent Implemented and
private “second-chance” Fast swap
working today
clean-page cache!! “device”!! (Linux + Xen)
“cleancache” “frontswap” Now in 2.6.32
shared server-side cluster inter-guest shared patch
filesystem cache memory?
“shared Under
cleancache” investigation
eph-em-er-al, adj., … transitory, existing only briefly, short-lived (i.e. NOT persistent)
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
25. Agenda
• Motivation and Challenge
• BRIEF Overview of Physical Memory Management
• BRIEF Transcendent Memory (“Tmem”) Overview
• Tmem Progress since Xen Summit 2009
• Self-ballooning + Tmem Performance Analysis
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
26. Transcendent Memory
update since Xen Summit 2009 (Feb’09)
• Tmem support officially released in Xen 4.0.0
• Xen boot option: tmem
• Enterprise-quality concurrency
• Complete save/restore and live migration support
• Page deduplication support (post-4.0.0)
• Linux-side patches posted, including
• ocfs2, btrfs, ext4 filesystem support (was ext3 only)
• sysfs support for in-guest tmem statistics
• Other tmem releases:
• Oracle VM 2.2 (10/2009)
• OpenSuSE 11.2 (11/2009); SLE11 SP1 later this year
• Oracle Enterprise Linux 5 update 4/5 rpm’s available soon
• targeting upstream Linux 2.6.35 (aka cleancache and frontswap)
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
27. Tmem page deduplication
• Now in xen-unstable (for 4.1) and xen-4.0-testing soon (for 4.0.1)
• Xen boot option: tmem_dedup
• Similar in intent to “page sharing” but
• very different in implementation
• completely transparent to tmem-enabled guests
• neither copy-on-write nor host swapping required
• all tmem ephemeral pages automatically are sharing candidates
• Can optionally be combined with:
• compression (tmem_compress); or
• “trailing zero elimination” (tmem_tze)
• Statistics available through existing “xm tmem-list”
• e.g., dedup+compression increase time per tmem op by ~10x
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
28. Known outstanding issue
Fragmentation! (ominous music plays here…)
• Xen page allocator allows “order>0” allocations
• order==1 2 contiguous pages, order==2 4 contiguous pages, etc.
• Some Xen features depend on order>0 allocations
• some are resilient and fall back to multiple one-page allocations
• some fail badly if not available (e.g., shadow code, domain creation)
• Ballooning+tmem quickly fragments all Xen RAM
• Hacky workaround in place for 4.0, but…
CALL-TO-ACTION:
Post-dom0-boot multi-page allocations are fundamentally
incompatible with all dynamic memory technologies and must
either be made resilient or eliminated from Xen!
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
29. Agenda
• Motivation and Challenge
• BRIEF Overview of Physical Memory Management
• BRIEF Transcendent Memory (“Tmem”) Overview
• Tmem Progress since Xen Summit 2009
• Self-ballooning + Tmem Performance Analysis
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
30. Test workload (overcommitted!)
• Dual core (Conroe) processor, 2GB RAM, IDE disk
• Four single vcpu PV VMs, in-kernel self-ballooning+tmem
• Oracle Enterprise Linux 5 update 4; two 32-bit + two 64-bit
• mem=384MB (maxmem=512MB)… total = 1.5GB (2GB maxmem)
• virtual block device is tap:aio (file contains 3 LVM partitions: ext3+ext3+swap)
• Each VM waits for all VMs to be ready, then simultaneously
• two Linux kernel compiles (2.6.32 source), then force crash:
• make clean; make –j8; make clean; make –j8
• echo c > /proc/sysrq-trigger
• Dom0: 256MB fixed, 2 vcpus
• automatically launches all domains
• checks every 60s, waiting for all to be crashed
• saves away statistics, then reboots
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
31. Measurement methodology
• Four statistics measured for each run
• Temporal: (1) wallclock time to completion; (2) total vcpu including dom0
• Disk access: vbd sectors (3) read and (4) written
• Test workload run five times for each configuration
• high and low sample of each statistic discarded
• use average of middle three samples for “single-value” statistic
• Five different configurations:
Features Self- Tmem Page Compression
enabled ballooning Dedup
Configuration
Unchanged NO NO NO NO
Self-ballooning YES NO NO NO
Tmem YES YES NO NO
Tmem w/dedup YES YES YES NO
Tmem w/dedup+ comp YES YES YES YES
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
32. Self-ballooning recap (see Xen Summit 2008)
• Goal: Allow memory overcommit
• Use in-guest feedback to resize balloon
• aggressively
• frequently
• independently
• configurably
• For Linux, size to maximum of:
guest • /proc/meminfo “CommittedAS”
• memory floor enforced by Xen balloon driver
• Userland daemon or patched kernel
Committed_AS: An estimate of how much RAM you would need to make a 99.99%
guarantee that there never is OOM (out of memory) for this workload. Normally the kernel
will overcommit memory. The Committed_AS is a guesstimate of how much RAM/swap you
would need worst-case. (From http://www.redhat.com/advice/tips/meminfo.html)
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
33. Unchanged vs. Self-ballooning only
Temporal stats
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
34. Unchanged vs. Self-ballooning only
Virtual block device stats
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
35. *Sigh* Why is there a performance hit?
Aggressive ballooning (by itself) doesn’t work very well!
• (Self-)ballooning indiscriminately shrinks the guest OS’s
page cache, causing refaults!
• Insufficiently responsive (self-)ballooning when guest OS
needs memory now, results in swapping… or OOMs!
PERFORMANCE WILL GET WORSE WHEN LARGE-
MEMORY GUESTS ARE AGGRESSIVELY BALLOONED
guest
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
36. Self-ballooning AND Transcendent Memory
…go together like a horse and carriage
• Self-ballooned memory is returned
fallow
to Xen and absorbed by tmem
fallow
• Most tmem memory can be
guest instantly reclaimed when needed
for a memory-needy or new guest
guest • Tmem also provides a safety valve
guest
when ballooning is not fast enough
fallow
Transcendent
guest memory
pool
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
37. Self-ballooning AND Tmem
Temporal stats
79% utilization*
5%-8% faster completion 72% utilization*
* 2 cores
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
38. Self-ballooning AND Tmem
virtual block device stats
31-52% reduction in sectors read
(no significant change in sectors written)
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
39. WOW! Why is tmem so good?
• Tmem-enabled guests statistically multiplex one shared
virtual page cache to reduce disk refaults!
• 252068 page (984MB) max (NOTE: actual tmem measurement)
• Deduplication and compression together transparently
QUADRUPLE apparent size of this virtual page cache!
• 953166 page (3723MB) max (actually measured by tmem… on 2GB system!)
• Swapping-to-disk (e.g. due to insufficiently responsive ballooning)
is converted to in-memory copies and statistically multiplexed
• 82MB at workload completion, 319MB combined max (actual measurement)
• uses compression but not deduplication
• CPU “costs” entirely hidden by increased CPU utilization
RESULTS MAY BE EVEN BETTER WHEN
WORKLOAD IS TEMPORALLY DISTRIBUTED/SPARSE
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
40. Transcendent Memory Update
Summary
Tmem advantages:
• greatly increased memory utilization/flexibility
• dramatic reduction in I/O bandwidth requirements
• more effective CPU utilization
• faster completion of (some?) workloads
Tmem disadvantages:
• tmem-modified kernel required (so only Linux now)
• higher power consumption due to higher CPU utilization
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
41. Transcendent Memory Update
Acknowledgements
Special thanks to Jeremy Fitzhardinge for many
reviews and improvements in the Linux-side tmem
code!
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer
42. For more information
http://oss.oracle.com/projects/tmem
or xen-unstable.hg/docs/misc/tmem-internals.html
dan.magenheimer@oracle.com
Update on Transcendent Memory on Xen (Xen Summit 2010) - Dan Magenheimer