Software Defined Storage For Dummies（Chap3-1）

Software Defined Storage For Dummies（Chap3-1）

chap3：Digging Deeper into IBM GPFS

##生词、短语

生词/短语	读音	释义
full-featured		adj. 全功能的；功能全面的
demanding	/dɪ'mɑːndɪŋ/	adj. 苛求的；要求高的；吃力的
gene sequencing	/'siːkwənsɪŋ/	基因测序
retail	/'riːteɪl/	n. 零售 adj. 零售的 vt. 零售；转述 adv. 以零售方式
biotechnology	/,baɪə(ʊ)tek'nɒlədʒɪ/	n. [生物] 生物技术；[生物] 生物工艺学
quotas	[k'wəʊtəz]	n. 配额（quota的复数）；[经管] 定额，[经] 限额
POSIX		Portable Operating System Interface,n. 可移植性操作系统接口
prohibitive	/prə(ʊ)'hɪbɪtɪv/	adj. 禁止的，禁止性的；抑制的；（费用，价格等）过高的；类同禁止的
viable	/'vaiəbl/	adj. 可行的；能养活的；能生育的
empower	/ɪm'paʊə; em-/	vt. 授权，允许；使能够

##基本知识点

• GPFS

Today it’s a full-featured set of file management tools, including advanced storage virtualization, integrated high availability, automated tiered storage management, and performance to effectively manage very large quantities of file data.

今天，它是一套功能全面的文件管理工具，包括高级存储虚拟化，集成高可用性，自动化分层存储管理以及有效管理大量文件数据的性能。

• **POSIX **

POSIX is an IEEE (Institute of Electrical and Electronics Engineers) family of standards for maintaining compatibility between different variations of UNIX and other operating systems.

POSIX是一个IEEE（电气与电子工程师协会）系列标准，用于维护UNIX和其他操作系统的不同变体之间的兼容性。

• 并行数据访问

GPFS accelerates time to results and maximizes utilization by providing parallel access to data. GPFS provides extreme performance and eliminates storage bottlenecks, by providing parallel access to data.

GPFS通过提供对数据的并行访问来加速实现结果的时间并最大限度地提高利用率。 通过提供对数据的并行访问，GPFS可提供极高的性能并消除存储瓶颈。

• GPFS achieves high-performance I/O by

Striping data across multiple disks attached to multiple servers,在连接到多个服务器的多个磁盘上分割数据
Providing efficient client side caching,提供高效的客户端缓存
Executing high-performance metadata (inode) scans,执行高性能元数据（inode）扫描
Supporting a wide range of file system block sizes to match I/O requirements ,支持各种文件系统块大小以匹配I / O需求
Utilizing advanced algorithms that improve I/O operations,利用改进I/O操作的高级算法
Using block-level locking based on a very sophisticated token management system to provide data consistency, while allowing multiple application servers concurrent access to the files
使用基于非常复杂的令牌管理系统的块级锁定来提供数据一致性，同时允许多个应用服务器并发访问文件。

• 文件完整性保护

When many servers need to use the same set of files at the same time, the file system needs to ensure that all the files are protected, so one server can’t change a file without the other servers knowing about the change. Keeping thousands of servers “in the loop” on file status is difficult and scaling up is even harder.

当许多服务器需要同时使用同一组文件时，文件系统需要确保所有文件都受到保护，因此一台服务器无法更改文件，而其他服务器却不知道这些更改。保持成千上万的服务器在文件状态“循环”是困难的，并且扩大规模更加困难。

---

GPFS provides file integrity protection through a token process that keeps file data consistent by always ensuring there is only one owner for any given file.

GPFS通过令牌处理提供文件完整性保护，通过始终确保给定文件只有一个所有者来保持文件数据的一致性。

There are two parts to managing tokens and file consistency: handing out the tokens and keeping file metadata up to date.

token manager: The server(s) that initially have the token for all files that are not in use is called the token manager. 最初拥有所有未使用文件令牌的服务器称为令牌管理器。

---

Multiple token managers help each other out by sharing the workload and by taking over when a fellow token manager fails.

多个令牌管理器通过共享工作负荷并在同伴令牌管理器发生故障时接管彼此。

---

When a file is opened, the token manager hands off the token for that file to the server that’s opening the file.

打开文件时，令牌管理器将该文件的令牌移交给正在打开该文件的服务器。

---

The server using the file is now responsible for all metadata changes to that file. If a server wants to open a file that is already open on another server, the token manager redirects the request to the server that already has the file open and lets the two servers work out the details among themselves.

使用该文件的服务器现在负责对该文件进行的所有元数据更改。如果服务器想要打开已经在另一台服务器上打开的文件，则令牌管理器会将请求重定向到已打开文件的服务器，并让两台服务器相互处理细节。

---

This sharing of metadata maintenance across the entire cluster is what makes GPFS scale very effectively.

在整个集群中共享元数据维护是使GPFS非常有效地扩展的原因。

• Simplifying Data Management

The global namespace is easy to administer and can be scaled quickly, as desired, by simply adding more scale-out resources — eliminating “filer-sprawl” and its associated issues.

全局命名空间易于管理，并可根据需要快速扩展，只需添加更多的扩展资源 - 消除“文件蔓延”及其相关问题。

---

A single GPFS command can perform a file system function across the entire cluster, and most can be issued from any server in the cluster. Optionally, you can designate a group of administration servers that can be used to perform all cluster administration tasks, or only authorize a single login session to perform admin commands cluster-wide. This allows for higher security by reducing the scope of server-to-server administrative access.

单个GPFS命令可以在整个集群中执行文件系统功能，并且大多数命令都可以从集群中的任何服务器发出。或者，您可以指定一组可用于执行所有群集管理任务的管理服务器，或者只授权一次登录会话以在群集范围内执行管理命令。这可以通过减少服务器到服务器管理访问的范围来提高安全性。

---

You can use snapshots to protect data from human error.

A snapshot is used to preserve the file system’s contents at a single point in time. It contains a copy of only the file system data that has changed since the last snapshot was created and keeps that data in the same pool as the original file, which keeps space usage at a minimum.

快照用于在单个时间点保留文件系统的内容。它仅包含自上次创建快照后发生更改的文件系统数据的副本，并将该数据与原始文件保存在同一个池中，从而将空间使用量保持在最低限度。

快照的作用

Snapshots provide an online backup capability that allows you (or an end user) to easily recover from an accidental file deletion, or the ability to compare a file to an older version.

---

Clustered NFS

To better enable end user access to a GPFS file system, the file system can be exported to clients outside the cluster through NFS (Network File System), including the capability of exporting the same data from multiple servers. This GPFS feature is called Clustered NFS (cNFS). Clustered NFS allows you to provide scalable file service with simultaneous access to a common set of data from multiple servers. The cNFS feature includes failover capability, so if a NFS server fails, the clients connected to that server automatically connect to another server in the cluster.

为了更好地支持最终用户访问GPFS文件系统，可以通过NFS（网络文件系统）将文件系统导出到群集外的客户端，其中包括从多个服务器导出相同数据的功能。此GPFS功能称为cluster NFS（cNFS）。cNFS允许您提供可伸缩的文件服务，并同时访问来自多个服务器的通用数据集。cNFS功能包括故障转移功能，因此如果NFS服务器发生故障，连接到该服务器的客户端会自动连接到群集中的另一台服务器。

NFS

NFS is a Network file system protocol that enables access to storage by using a standard protocol over a TCP/IP network. NFS protocol access is commonly provided by a networkattached storage (NAS) appliance or similar device. Samba enables file and print services for Microsoft Windows clients from UNIX and Linux based servers.

NFS是一种网络文件系统协议，可通过TCP/IP网络上的标准协议访问存储。NFS协议访问通常由网络附加存储（NAS）设备或类似设备提供。Samba为来自基于UNIX和Linux的服务器上的Microsoft Windows客户端启用文件和打印服务。

• Cluster Configurations

   1. Shared disk

A shared disk cluster is the most basic environment. In this configuration, the storage is directly attached to all servers in the cluster. Application data flows over the SAN, and control information flows among the GPFS servers in the cluster over a TCP/IP network.

共享磁盘集群是最基本的环境。在这个配置中，存储直接连接到集群中的所有服务器。应用数据在SAN上流动，并且控制信息在TCP/IP网络中在集群中的GPFS服务器之间流动。

This configuration is best for small clusters (1 to 50 servers) when all servers in the cluster need the highest performance access to the data. For example, this configuration is good for high-speed data access for digital media applications or a storage infrastructure for data analytics.

   2. Network Shared Disk (NSD) protocol

GPFS uses a network to transfer control information and data to NSD clients. The network doesn’t need to be dedicated to GPFS, but it should provide sufficient bandwidth to meet your GPFS and other applications sharing the bandwidth.

GPFS使用网络将控制信息和数据传输到NSD客户端。 网络不需要专用于GPFS，但它应该提供足够的带宽来满足您的GPFS和其他共享带宽的应用程序。

---

In a NSD server configuration, a subset of the total server population is defined as NSD servers. The NSD servers are responsible for the abstraction of disk data blocks across an IP-based network. The fact that I/O is remote is transparent to the application. Figure 3-3 shows an example of a configuration where a set of compute servers are connected to a set of NSD servers via a high-speed interconnect or an IP-based network (such as Ethernet). In this example, data to the NSD servers flows over the SAN, and data and control information flows to the clients across the LAN.

在NSD服务器配置中，服务器总体的一个子集被定义为NSD服务器。NSD服务器负责在基于IP的网络上抽象磁盘数据块。I/O是远程的这一事实对应用程序是透明的。图3-3显示了一组配置示例，其中一组计算服务器通过高速互连或基于IP的网络（如以太网）连接到一组NSD服务器。在本例中，NSD服务器的数据流经SAN，数据和控制信息通过LAN流向客户端。

An NSD server architecture is well suited to clusters with sufficient network bandwidth between the I/O servers and the clients. For example, statistical applications like financial fraud detection, supply chain management, or data mining.

   3. Empowering global collaboration

GPFS provides low latency access to data from anywhere in the world with Active File Management (AFM) distributed disk caching technology. AFM expands the GPFS global namespace across geographical distances, providing fast read and write performance with automated namespace management from anywhere in the world. As data is written or modified at one location, all other locations get the same data with minimal delays. These game-changing capabilities accelerate project schedules and improve productivity for globally distributed teams.

采用活动文件管理（AFM）分布式磁盘缓存技术，GPFS可以从世界任何地方提供低延迟访问数据。AFM跨越地理距离扩展了GPFS全局名称空间，从世界上任何地方通过自动名称空间管理提供快速读写性能。随着数据在一个位置被写入或修改，所有其他位置以最小的延迟获取相同的数据。这些改变游戏规则的功能可加快项目进度并提高全球分布的团队的生产力。