Asynchronous Replication with NexentaStor

本文深入探讨了 NexentaStor 的异步复制选项,包括 AutoTier 和 AutoSync 的工作原理、调度器配置、文件选择、目标系统指定、双向复制使用场景及监控方法。

摘要生成于 C知道 ,由 DeepSeek-R1 满血版支持, 前往体验 >

Asynchronous Replication with NexentaStor

This document discusses variousasynchronous replication options with NexentaStor. NexentaStor provides twodifferent replication services, AutoTier and AutoSync.

Both AutoSync and AutoTier supportincremental data transfers. After the initial replication, only changed data onthe source is sent for subsequent transfers.

AutoTier

AutoTier typically uses rsync as the data transferprotocol. Typical rsync options used by AutoTier are “-HlptgoD”.

AutoSync

The AutoSync service requires that bothsource and target are ZFS-based systems. Depending on the options used, theremay be further restrictions. For example, the “RR” protocol is supported onlybetween two NexentaStor systems.

Scheduler

NexentaStor includes a sophisticatedscheduler with its replication service. Replication can be scheduled for aspecific time, and run once or periodically. Some examples are:

o  Run at 3am

o  Run every 4 hours

o  Run once a week

o  Run right now

o  Run once, at 5pm.

The smallest time interval you canconfigure in NMV is 1 minute. With NMC you can configure AutoSync to run indaemon mode and do transfers every few seconds.

Controlling which files are replicated

NexentaStor provides the ability to specifyfiles to not be included in the replication. This can be indicated whencreating or editing the replication service. To do this select More Options anduse the Exclude folders option. Here you specify patterns of files that you donot want to be included in the data transfer. For example, here is how youwould specify to not send files ending in “.jsp”.


For multiple patterns, please type as    *.jsp\|*.class\|*.properties

Specifying the Target System

NexentaStor can provide replication services to and from a variety of storage systems. However not all

Bi-directional Replication

NexentaStor can allow for replication to beperformed bi-directionally. A separate replication serviceis configured for each direction.

For bi-directional replication you need touse the AutoTier service.

Various usage models for bi-directionalreplication might be:

o  Add/update files only if theyare newer than the receiver

o  Add/update files only if theydon’t already exist on the receiver

o  Add/update files only if they exist on the receiver

Add/update files only if they are newer than receiver

For this use case both the source andtarget storage systems are updating files in the same file systems. To resolveconflicts if the file is edited in both places, the file changed most recentlyis kept. Thus we will rely on the file modification times and it is criticalthat the source and target systems be using a time service such as NTP.

We need to use AutoTier for this use case because AutoSync makes certainassumptions about the target directories not being in use or modified.But AutoTier by default assumes that the source is the master and thereis no other unique activity happening on the target. So for this use casewe need to make some changes to the default rsync command-line options. To editthe rsync parameters, you need to select More Options when configuring theservice. This will cause the Window of available options to expand, as shown inthe following screenshot.

We need to make the following changes tothe default rsync options:


Remove:

--delete: If we use this option, then anyadditional files created by the receiver will be deleted.

--inplace: this doesn’t seem to be a goodoption if the receiver might be editing the files

Add:

--update: this will skip replication forfiles that are newer on the receiver

--times: preserve modification times. Thisoption may not be necessary as they seem to already be preserved.

When configuring the service, choose“Replicate content”. Select the source folder.

For the folder on the target system,specify the wildcard at the end of the target directory. For example, if thetarget is a remote system then you might reference the folder like this:“/mytarget/*”. If you don’t use the “Replicate content”option you might end upwith additional directories getting created.

For receiving data from the target system,you need to again specify the wildcard for the receiving directory. Howeverthis can’t be done using NMV. Instead, create the service first in NMC using“setup auto-tier”. This will allow you to specify the wildcard for the targetfolder on the local system. After the service is created in NMC, you can thenedit the rsync options using NMV as shown above.

Monitoring

The replication service is an importantmechanism for data protection. It is important to be able to check anddetermine if the service is functioning correctly. This can be done easily fromNMV. On the summary view for the services, there is an option to View Log:


After selecting View log, you may be askedto provided the administrative password if you are not already logged in. Nexta dialog box appears showing you the log of past executions of the replicationservice and whether the task was successful.

 

Note that by default NexentaStor willre-run the service automatically if the service failed due to a loss ofconnectivity. This can be changed by selecting “More Options” when creating orediting the service.

 

Related Topics

AutoCDP provides synchronous replication.The user manual can be found at http://www.nexenta.com/corp/documentation/product-documentation.

The rsync protocol options are describedhere: http://rsync.samba.org/ftp/rsync/rsync.html.

内容概要:《中文大模型基准测评2025年上半年报告》由SuperCLUE团队发布,详细评估了2025年上半年中文大模型的发展状况。报告涵盖了大模型的关键进展、国内外大模型全景图及差距、专项测评基准介绍等。通过SuperCLUE基准,对45个国内外代表性大模型进行了六大任务(数学推理、科学推理、代码生成、智能体Agent、精确指令遵循、幻觉控制)的综合测评。结果显示,海外模型如o3、o4-mini(high)在推理任务上表现突出,而国内模型如Doubao-Seed-1.6-thinking-250715在智能体Agent和幻觉控制任务上表现出色。此外,报告还分析了模型性价比、效能区间分布,并对代表性模型如Doubao-Seed-1.6-thinking-250715、DeepSeek-R1-0528、GLM-4.5等进行了详细介绍。整体来看,国内大模型在特定任务上已接近国际顶尖水平,但在综合推理能力上仍有提升空间。 适用人群:对大模型技术感兴趣的科研人员、工程师、产品经理及投资者。 使用场景及目标:①了解2025年上半年中文大模型的发展现状与趋势;②评估国内外大模型在不同任务上的表现差异;③为技术选型和性能优化提供参考依据。 其他说明:报告提供了详细的测评方法、评分标准及结果分析,确保评估的科学性和公正性。此外,SuperCLUE团队还发布了多个专项测评基准,涵盖多模态、文本、推理等多个领域,为业界提供全面的测评服务。
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值