Several problems with Ceph optimization

Source: Internet
Author: User

Ceph Cluster Problem Grooming
    • Created: Linhaifeng, last modified: Yesterday 5:35 afternoon

1: Will the data be sent to the log disk and deleted immediately or delayed?

Validation: Tuning Parameters

#从日志到数据盘最大同步间隔间隔秒数, Default: 5

Filestore Max sync interval = 15

Process Analysis: A client object sent to Ceph's PG, note that the log is written to return the results, and then within 15 seconds of the interval,

Three OSD node data is synchronized to delete the data in the log disk, while notifying the client can delete their own in-memory data,

Supports the ability of clients to resubmit data based on their own in-memory cache if data is lost at this time

650) this.width=650; "class=" Confluence-embedded-image "src=" http://wiki.ky.com/download/attachments/3415901/ Image2016-5-25%2017%3a18%3a33.png?version=1&modificationdate=1464167917000&api=v2 "alt=" image2016-5-25% 2017%3a18%3a33.png?versio "/>

Issue 2: After multiple objects to a primary OSD, is it concurrency or blocking

Verification: Because the client reads and writes are directly and primary OSD dealings, and we use the XFS file system is characterized by the write log is returned, so we use SSD log disk, to quickly handle a large number of concurrent write requests

Question 3: How big should each partition be by using an SSD as a log disk?

Official recommendation (one * maximum sync interval seconds for hard disk and network bandwidth) * *

Maximum bandwidth of this project 12Gbps maximum synchronization interval of 15 seconds

An sshd is divided into two partitions, one log disk per partition, and each log disk size (size per partition) =12gbps*15*2=360gb=45g

Log disk:

Solid-state Drives

Less partitioning (partition alignment, partitioning by script using percentages)

Small partition (do not fill the entire hard disk partition, reserved part of the space, we sshd is 200G, can use about 150G), the usage rate should not exceed 90%, do not defragment.

Problem 4:file split into single object performance optimal

From the performance point of view should be set to a multiple of 2, if it is small files can be used in the majority of 4

If it is a mirror, or a disk image file for a cloud host, you can use 8

Units are all m by default


This article is from "A Good person" blog, please be sure to keep this source http://egon09.blog.51cto.com/9161406/1783318

Several problems with Ceph optimization

Contact Us

The content source of this page is from Internet, which doesn't represent Alibaba Cloud's opinion; products and services mentioned on that page don't have any relationship with Alibaba Cloud. If the content of the page makes you feel confusing, please write us an email, we will handle the problem within 5 days after receiving your email.

If you find any instances of plagiarism from the community, please send an email to: info-contact@alibabacloud.com and provide relevant evidence. A staff member will contact you within 5 working days.

A Free Trial That Lets You Build Big!

Start building with 50+ products and up to 12 months usage for Elastic Compute Service

  • Sales Support

    1 on 1 presale consultation

  • After-Sales Support

    24/7 Technical Support 6 Free Tickets per Quarter Faster Response

  • Alibaba Cloud offers highly flexible support services tailored to meet your exact needs.