How to build a private storage cloud

企业构建内部云存储时必须考虑弹性,选择正确的平台,并允许工作流,堆栈部署和公共云集成。

每个云存储选项都有其优点和缺点。企业需要根据自己的具体需求,规模大小,以及资金预算来选择采用哪种云存储,重要的是权衡所有云和内部部署选项。可以下载一些综合指南,其中专家分析和评估当前可用的每个云存储选项,以便企业可以决定采用哪个云计算模式-公共云,私有云,或混合云。

企业如何去构建自己的私有存储云?首先,让我们回顾一下云计算的真正含义。云计算的标准定义包括以下特点:弹性增长和缩减消耗资源;交付即服务,以抽象术语而不是在物理硬件上定义的服务产品的标准集合;多租户支持多客户端;按需访问很少或不需要手动干预的资源;报告和计费,以及基于时长收费的详细报告。

私有存储云应该反映这些相同的功能。在这种情况下的客户,应该能够请求存储容量,而不必担心如何交付。因此,已经使用多年并且可能专注于物理技术(例如HDD速度或HDD/闪存)的服务目录需要刷新,以更加专注于服务度量。这意味着使用I/O密度(每千兆字节存储的IOPS),延迟,吞吐量,数据可用性和弹性等术语。

多租户是指安全和性能隔离。安全性确保数据在私有存储云用户之间不可见,而诸如服务质量(QoS)的性能特征使得每个用户接收一致的服务等级,而不管其系统负载如何。按需访问保证了客户对资源的请求,尤其是IT存储管理员可以更少地进行干预。报告必须处理更精细的存储利用率,包括报告各个团队或业务领域的能力。

创造弹性

企业的私有存储云计划列表中的第一个项目就是弹性,这有两种情况:第一,从客户的需求扩展和收缩使用的能力,第二,为系统管理员能够部署更多的基础设施作为需求保证。虽然有些担心部署的硬件不会使用,如果最终用户可以轻松地给存储,但这很少发生。

这里的挑战是通过将新硬件部署到数据中心,同时管理技术更新周期,而不影响应用程序可用性,继续满足需求。实现即时部署新硬件对于大多数IT部门来说是一种艺术和科学,大多数IT部门不像亚马逊和微软公司这样规模的行业巨头拥有大量的现金和人力资源,他们必须通过预测何时需要证明新的硬件采集来提供无限的资源。

IT部门进行需求预测,需要与业务部门接触,以规划潜在的未来项目及其存储需求。如果IT部门能够深入了解未来可能的存储资源需求,这些需求可以更容易规划,特别是如果它们是非核心产品,如对象或高性能存储。

科学是对充分的存储增长信息的发现。许多IT环境使用精简配置,这意味着随着数据写入分配的空间,其物理存储容量会随时间增加。并且因为对计划的存储消耗的保留,很少被立即完全利用(例如,1TB请求可能仅使用一段时间(例如50GB一天),并且被确定为三年的增长),文件系统和对象存储将自然地增加利用率,而应用程序写入更多的数据。这使得IT部门必须具有准确和详细的工具来测量随时间增长的存储消耗,同时使用该数据来产生有意义的增长预测。

此外,确定何时部署新硬件需要了解和管理供应商潜在客户,硬件部署和配置时间。在企业中,IT部门仍然拥有这些问题,当企业采用公共云存储时,这些问题当然在云计算服务提供商(CSP)的管辖范围内。

选择平台

拥有正确的存储平台是高效部署新硬件的关键。横向扩展作为拓展容量技术的选择可以使新的部署相对简单,因为您企业需向现有配置添加硬件以增加容量。

非法闯入是指黑客利用企业网络的安全漏洞,不经允许非法访问企业内部网络或数据资源,删除、复制甚至毁坏数据。

招标办加强保密工作 维护信息安全

大多数现代对象和块横向扩展产品执行一定程度的重新平衡,重新分配数据以利用新的容量,并从硬件获得最大的性能。由于可扩展性限制,单片放大架构可能难以管理,而旧的传统存储系统可能不会顺理成章地负载平衡以利用新的物理容量。这意味着传统架构必须更仔细地设计,以便在物理硬件上负载平衡逻辑资源的分布。这些平台中的许多都具有用于在存储平台内移动LUN的工具,从而减轻了一些平衡问题。

潞城供电组织专线用户培训 做足春检保安全工作

多租户和服务质量(QoS)成为在为私有存储云选择存储平台时需要考虑的关键功能。看看移动服务提供商(CSP)提供的服务指标,人们看到性能是根据IOPS和吞吐量评分,有些提到I/O延迟。无论移动服务提供商(CSP)是否正在运行完全加载,这些服务级别被认为对用户可用,并不是传统存储。因此,服务质量变得非常重要,作为确保最终用户获得他们需要的性能,或将性能限制到他们用新系统支付的水平的工具。

需要应用程序编程接口(API)


近年来,存储设备世界已经看到了微小的管理演变。企业通常通过GUI和一些命令行界面(CLI)交互手动管理存储,使用“提交”阶段进行更改。CLI使存储管理员能够对配置和停用过程进行脚本化,从而实现一定程度的自动化。但是,创建脚本是一个耗时的过程。多年来,供应商转向实现使存储可编程的API,通过授权的API调用来设置配置。配置数据也可以很容易地提取,一些存储平台生成非常详细的详细指标。

应用程序编程接口

应用程序编程接口(API)改变了企业存储管理的方式。将来,API将驱动自动化,并从存储配置中删除大多数人工干预,使私有云存储广泛应用于更多的企业。

API还支持自动化,使“工作人员”脱离配置存储的过程。现在,存储可以通过一个或两个API调用映射到主机。一些平台本地实现API,而一些平台已经围绕现有API工具构建了API封装。这里的关键要求是确保API,CLI和GUI操作顺畅,而不是相互跨越。

工作流很重要

提供私有云存储的最后一个难题是执行某种工作流程。用户请求必须以允许请求被验证然后实现的方式被服务。公共云通过用户提供信用卡或用于计费的其他支付方法来实现这种验证过程。之后,通过Web门户或API配置服务。在企业中,请求存储的传统过程是具有手动管理请求的内部进程,基于服务票证中的详细信息向主机配置存储。机票的所有者负责确保允许业务部门“购买”正在订购的存储,然后负责履行。

即收即付

企业可以使用信用卡购买公共云资源,并以信用卡结算。工作流程中的这一变化,意味着许多组织将需要在部署内部云存储时考虑实施计费和扣款。

在私有云中,目的是使这个过程尽可能实现自动化。例如,EMC公司的ViPR等工具允许用户围绕存储自动化构建工作流程。日立数据系统公司提供Hitachi自动目录来构建有关存储和其他资源配置的工作流程。

许多组织将需要考虑私有存储云可以引入的计费变化。如果未实施结算和扣款,则没有问题,因为IT部门将继续承担提供服务的成本,并且可能会继续按项目收费。然而,如果必须支付新的资源,则可能需要引入对金融做法的一些改变(其可能当前包括直接为硬件支付的),以允许业务单元的基于服务的计费来支付成本。

堆栈部署

比存储团队更广泛,企业可以将存储自动化构建到私有云框架(例如OpenStack)中,以便尽可能地减少配置。初始OpenStack部署没有持久存储功能,因此,一些项目被实例化以管理外部存储阵列的集成。最终的OpenStack公司Cinder项目处理块存储并自动将LUN映射到OpenStack实例,而Manila则提供文件系统数据的集成,Swift为对象存储提供了一个API。

更广泛的堆栈

云存储,无论是内部部署还是公共云,将构成更广泛的基础设施堆栈的一部分。这意味着与诸如OpenStack或vCloudDirector之类的平台集成。

同时,存储供应商可以编写插件,使OpenStack框架能够按需配置和映射存储LUN。许多硬件和软件公司已经支持所有的OpenStack存储API。Cinder项目支持矩阵列出了OpenStack平台的每个版本中支持的供应商功能。

公共云的集成

展望未来,云计算的世界不会完全是采用公共云或私有云,而是两者的混合。因此,将需要在公共云和私有基础设施之间移动数据和应用程序,后者提供额外的数据保护(备份)和增加的可用性。企业还可以使用公共云存储来进行工作负载突发和归档。

用于在内部部署数据中心和公共云位置之间移动应用程序和数据的产品已经开始上市。诸如Cloudian(HyperStore)和Hitachi数据系统(HitachiContentPlatform)之类的对象存储供应商提供了将内部数据存档到云计算中的能力,同时极大地保留了跨所有内容进行搜索的能力,就像在单个视图中一样。

对于数据保护,Druva公司和Zerto公司都提供允许用户备份和恢复公共云中的本地虚拟机(VM)的产品。虚拟机映像的转换和其他驱动程序的注入由软件处理,作为备份和迁移过程的一部分。

虚拟化

在运行服务器虚拟化的平台中,存储不常映射到物理主机。创建虚拟机实例存储的大多数工作由管理程序管理软件处理。VMware通过vRealizeAutomation和vCloudDirector提供自动化,而微软公司提供SystemCenter2016。

Velostrata公司进一步允许在公共云中启动虚拟机以防云爆发。这可以用于在具有比现场可用的资源容量更大的资源容量的虚拟机上运行应用程序,或者将工作负载移动到公共云以应对增加的需求。一旦需求激增,虚拟机可以返回现场。

同时,虚拟化供应商开始与云供应商合作,以便于将应用程序迁移到公共云中。例如,VMware公司最近宣布在AWS云计算基础上推出VMwareCloud,并与IBM公司建立了合作伙伴关系。它还引入了跨云架构作为管理多个云部署的一种方式。MicrosoftAzureStack使春Azure功能可以在私有数据中心运行,并链接到Azure公共云。

原告广州农村商业银行股份有限公司诉被告篮丽娜、卢银坤、篮俊…
很显然,如今与公共云实现不同的是,主要在自动化程度的地方,以充分利用私有云存储。作为最关键的一部分,工作流可能还不成熟,需要在私有云进一步发展。

这里的一部分挑战是改变内部业务团队的行为。这是公共云帮助促进的,也应该被采纳为内部资源的交付模式。
该文章作者已设置需关注才可以留言
微信扫一扫关注该公众号
When an enterprise builds an internal cloud storage, it must consider the flexibility, choose the right platform, and allow the integration of workflow, stack deployment and public cloud.
Each cloud storage option has its advantages and disadvantages. Companies need to choose according to their specific needs, size, and budget to choose what kind of cloud storage, it is important to weigh all the cloud and internal deployment options. You can download a number of comprehensive guides, where experts analyze and evaluate the currently available cloud storage options, so that companies can decide which cloud computing model to use – public cloud, private cloud, or hybrid cloud.
How do companies build their own private storage cloud? First, let’s look at the real meaning of cloud computing. The standard definition of cloud computing includes the following features: elastic growth and reduce the consumption of resources; delivery service, in abstract terms not defined in the physical hardware service product standards set; multi tenant support multi client; on-demand access requires little or no manual intervention resources; reporting and billing, and based on a detailed report long fee.
The private storage cloud should reflect these same features. In this case, the customer should be able to request storage capacity without having to worry about how to deliver. As a result, service directories that have been used for many years and may focus on physical technologies (such as HDD speed or HDD\/ flash) need to be refreshed to focus more on service metrics. This means that the use of I\/O density (IOPS per Gigabyte storage), latency, throughput, data availability and flexibility, etc..
Multi tenant refers to security and performance isolation. Security ensures that data is not visible between private storage cloud users, and the performance characteristics such as quality of service (QoS) allow each user to receive a consistent level of service regardless of the system load. On demand access guarantees the customer’s request for resources, especially the IT storage manager can intervene less. The report has to deal with finer storage utilization, including the ability to report to each team or business area.
Creating flexibility
The first project of enterprise private storage cloud list is elastic, there are two situations: first, the ability to use, from customer demand expansion and contraction of the second, to deploy more infrastructure as demand guarantee for system administrator. Although some worry about the deployment of hardware will not be used if the end user can easily be stored, but this rarely occurs.
The challenge here is to continue to meet the demand by deploying new hardware to the data center, while managing the technology update cycle without affecting the usability of the application. Realize the immediate deployment of new hardware for most IT departments is a kind of art and science, giant scale unlike Amazon and the Microsoft Corp the most IT industry sector has a lot of cash and human resources, they must go through to predict when the need to prove that the new hardware collection to provide unlimited resources.
IT Department of demand forecasting, need to contact with the business sector to plan for potential future projects and their storage needs. If the IT department is able to gain insight into possible future storage resource requirements, these requirements can be easier to plan, especially if they are non core products, such as objects or high-performance storage.
Science is the discovery of sufficient storage growth information. Many IT environments use reduced configuration, which means that the physical storage capacity increases with time as the data is written into the allocated space. And because of the consumption of the memory retention plan, is rarely immediately fully utilized (for example, 1TB requests may only be used for a period of time (e.g. 50GB day), and was determined to be three years of growth), file system and object storage will naturally increase the utilization rate and application to write more data. This allows the IT sector to have accurate and detailed tools to measure the amount of storage consumed over time, while using the data to generate meaningful growth forecasts.
In addition, it is necessary to understand and manage the potential customer, hardware deployment and configuration time when deploying new hardware. In the enterprise, IT departments still have these problems, when the enterprise uses public cloud storage, of course, these problems within the jurisdiction of the cloud computing service provider (CSP).
Selection platform
Having the right storage platform is the key to efficient deployment of new hardware. Horizontal expansion as an option to expand the capacity of the technology can make the new deployment is relatively simple, because your business needs to add hardware to the existing configuration to increase capacity.
Most modern object and block horizontal expansion products perform a certain degree of rebalancing, redistribute data to take advantage of new capacity, and derive maximum performance from hardware. Because of the scalability limitations of monolithic amplifier architecture can be difficult to manage, but the traditional storage system may not be logical to old load balancing by using the physical capacity of the new. This means that the traditional architecture must be designed more carefully so as to balance the distribution of logical resources on physical hardware. Many of these platforms have tools for moving LUN in a storage platform, thereby reducing some of the balance issues.
Multi tenant and quality of service (QoS) has become a key feature to consider when selecting a storage platform for a private storage cloud. Look at the service metrics provided by mobile service providers (CSP), people see performance is based on the IOPS and throughput score, some mention I\/O delay. Whether or not the mobile service provider (CSP) is running fully loaded, these service levels are considered to be available to the user, not the traditional storage. As a result, the quality of service becomes very important as a tool to ensure that end users get the performance they need, or to limit their performance to the level they pay with the new system.
Application programming interface (API)
In recent years, the world of storage devices has seen the evolution of micro management. The enterprise usually manages the storage manually by GUI and some command line interface (CLI) interactions, using the commit phase. CLI enables the storage manager to script the configuration and deactivation process to achieve a certain degree of automation. However, creating scripts is a time-consuming process. Over the years, the vendor has turned to enable the storage of programmable API, through the authorized API tone used to set the configuration. Configuration data can also be easily extracted, and some storage platforms generate very detailed detailed metrics.
Application programming interface
Application programming interface (API) changes the way enterprise storage management. In the future, API will drive automation, and remove most of the manual intervention from the storage configuration, so that private cloud storage is widely used in more enterprises.
API also supports automation to staff out of the process of configuration storage. Now, the storage can be mapped to the host via one or two API calls. Some platforms to achieve local API, and some platforms have been built around the existing API tools API package. The key here is to ensure that API, CLI and GUI operate smoothly, rather than cross each other.
Workflow is very important
The last difficulty in providing private cloud storage is the execution of a workflow. User requests must be served in a way that allows the request to be verified and implemented. The public cloud implements this verification process by providing credit cards or other payment methods for billing. After that, through the Web portal or API configuration services. In the enterprise, the traditional process request is stored with manual management request of internal processes, detailed information service based on the ticket to the host configuration storage. The owner of the ticket is responsible for ensuring that the business department is allowed to purchase the store that is being ordered, and then is responsible for fulfilling the.
Pay as you go
Companies can use credit cards to buy public cloud resources, and credit card settlement. This change in the workflow, means that many organizations will need to consider the implementation of billing and deduction in the deployment of internal cloud storage.
In a private cloud, the purpose is to automate the process as much as possible. For example, EMC’s ViPR and other tools allow users to build workflow around storage automation. Hitachi data systems provides the Hitachi directory to build workflow for storage and other resource configurations.

最新威胁报告称恶意软件的增速有所放缓,分析表明,黑客再使用更为复杂的攻击方式如高级可持续性威胁APT,防范APT,也得使用更为复杂和精准的多层防御体系。

猜您喜欢

新手指南:DVWA-1.9全级别教程(完结篇,附实例)之XSS
新发布“安全朗朗然”视频将促进国民安全意识
中国企业到海外,如何应对风险挑战:
邓超携《乘风破浪》登陆快本 现场释放表情包
WGRZ FOUNDATIONHORSES
建立信息安全培训计划