|

Linux Kernel 4.19.70 Release

This post summarizes new features, bugfixes and changes in Linux kernel release 4.19.70. Linux 4.19.70 Release contains 95 changes, patches or new features. In total, there are 101,521 lines of Linux source code changed/added in Linux 4.19.70 release compared to Linux 4.19 release. To view the source code of Linux 4.19.70 kernel release online, please…

What is the difference between work conserving I/O scheduler and non-work conserving I/O scheduler?

What is the difference between work conserving I/O scheduler and non-work conserving I/O scheduler? In a work-conserving mode, the scheduler must choose one of the pending requests, if any, to dispatch, even if the pending requests are far away from the current disk head position. The rationale for non-work-conserving schedulers, such as the anticipatory scheduler…

How to get processes’ I/O utilization percentage

Two notices: 1, a process has only one main thread which is itself. 2, a process has many threads. Solution 1: Please use taskstats [1] related interfaces, and send TASKSTATS_TYPE_PID and TASKSTATS_TYPE_TGID commands to kernel to get a process’s ‘blkio_delay_total’ parameter for a process with one main thread and a process with threads separately. Solution…

|

I/O Microscopy: Tasks’ Disk I/O Information with High Accuracy

Abstract Most popular task monitor systems (such as top, iotop, proc, etc) can only get tasks’ disk I/O information like tasks’ I/O utilization percentage every seconds due to kernel timer/tick frequency and high time cost of system interfaces. This article presents I/O Microscopy, a new way to get tasks’ disk I/O information with high accuracy….

| |

x-data-plane feature in QEMU/KVM

Abstract In systems, sometimes, we use one global lock to keep synchronization among different threads. This principle also happens in QEMU/KVM (http://wiki.qemu.org/Main_Page) system. However, this may cause lock contention problem. The performance/scalability of whole system will be decreased. In order to solve this problem in QEMU/KVM, x-data-plane feature is designed/implemented, which the high-level idea is…

|

How sched_min_granularity_ns, sched_latency_ns and sched_wakeup_granularity_ns in CFS affect the timeslice of processes

Abstract Currently, the most famous process scheduling algorithm in Linux Kernel is Completely Fair Scheduling (CFS) algorithm. The core idea of CFS is to let each process share the same proportional CPU resources to run so that it is fair to each process. In this article, I will introduce how sched_min_granularity_ns and sched_latency_ns work internal…

|

SSD Enabled For DreamHost Shared Hosting: Simple Performance Measurement

SSD is common for VPS and PaaS virtual machines for higher I/O performance. Now, it is coming to shared hosting too. DreamHost states that “Now with solid state drives (SSDs), our standard web hosting loads pages 200% faster”. We ourselves are happy to know this performance improvement with the price kept the same. Good work,…

| |

Managing Xen Dom0′s CPU and Memory

The performance of Xen’s Dom0 is important for the overall system. The disk and network drivers are running on Dom0. I/O intensive guests’ workloads may consume lots Dom0′s CPU cycles. The Linux kernel calculates various network related parameters based on the amount of memory at boot time. The kernel also allocate memory for storing memory…

|

Xen DomU’s I/O Performance of LVM and loopback Backed VBDs

This posts list benchmark (using bonnie++) result of I/O performance of Xen LVM and loopback backed VBDs. The configuration of machines Dom0 VCPU: 2 (Intel(R) Xeon(R) CPU E5520  @ 2.27GHz) Memory: 2GB Xen and Linux kernel: Xen 3.4.3 with Xenified 2.6.32.13 kernel DomU VCPU: 2 Memory: 2GB Linux kernel: Fedora (2.6.32.19-163.fc12.x86_64) DomU’s profile: name=”10.0.1.200″ vcps=2…