当涉及管理混合动力和多罩环境时，有许多选项但没有容易的路径也不缺乏挑战.
虽然云计算已经以某种形式出现超过十年，但是从私人，内部部门或公共场所管理其当前企业迭代的工具仍然以快速的速度发展。Gartner表示，超过90家供应商 - 包括IBM / Red Hat，VMware，Cloudbolt，Flexera，Scalr，Cisco和Nutanix - 提供不同程度的云管理功能。
虽然有许多选择，但是，组织努力管理多云环境，罗伊特·斯坦尔（云管理业务部）VMware副总裁Roy Ritthaler表示。
“With workloads deployed in multiple public clouds, multi-cloud Kubernetes, private cloud/data centers and edge locations, most organizations find it challenging to get a unified view of the health of their environments as well as manage costs, ensure security and improve operational governance while automating core processes,” Ritthaler said.
Ritthaler说，这不仅仅是一种技术挑战，而且还有一个人和流程挑战。“缺乏统一的供应工具，友好的操作知名度，缺乏整体性能和成本洞察力和互操作性和集成问题，均为静坐资源，分散的团队和管理工具扩散。涉及多个角色 - IT运营，Devops /开发人员，财务和业务（LOB）领导者，需要大量培训，协作和过程变更随机组织拥抱云模型。“
最近IDC研究发现大多数企业期望他们需要净新的多箱管理工具，以跟上其新兴的业务和基础设施 - 运营需求。
“Multicloud architectures are introducing a new wave of management complexity as developers and business groups implement cloud services and tools that best align with their application and business innovation road maps with limited regard for corporate preferences. The introduction of containers, microservices, and Kubernetes creates further complexity.” IDC stated.
Over the next two years, enterprise decision makers are expected to prioritize investments in analytics, performance monitoring and reporting, capacity optimization, cost management, as well as automation and self-service to augment management capabilities for multicloud and governance, according to IDC .
“这些管理工具深受互联。成本决定必须在容量要求和应用程序性能的背景下进行，“IDC表示。
还有关于应用显影密度的企业焦虑，跨越不同的云提供商。最近Enterprise Management Associates studysaid there are 2,316 Python libraries related to AWS, Azure, and Google Cloud that developers download approximately 13 million times per day to 112 different, mostly Linux-based, operating systems.
“虽然个别项目通常保持在单个云的界限内，但EMA也看到越来越多的云横跨多个云的项目。微服务的快速增长增加了这一趋势，同时强调，为开发人员和IT运营商的统一治理和管理层的迫切需要有助于优化释放效率和运行可靠性同时“Ema表示。
Such a wide variety of projects has caused many customers to look for help managing workloads across many environments, which requires multiple consoles and tools, IDC stated.
“As enterprises manage multicloud environments and the number of consoles and tools grows, it’s common to experience challenges that stem from siloed data—an inevitable, common side effect of migrating applications running on legacy systems to disparate cloud environments,” said Briana Frank, director of product management with IBM Cloud. “As enterprises move disconnected data from one cloud to another to be used by various applications, they often experience performance issues and significant cost increases up to 300%, according toIBM.research.”
客户抱怨使用多箱服务，说由于云之间的数据传输并增加了其人员配置，因此提高了成本。Douglas Gourlay，Arista的云网络软件副总裁兼总经理。
Getting a unified view among clouds
此外，由于企业迁移到多个云，他们很快就会发现每个云提供商都是唯一的, which adds challenges to manage those environments, “like network architectures, features, and scale, which creates a steep learning curve for customers to operate in the cloud and creates operational challenges across existing environments like data-center and campus networks,” Gourlay said.
在Arista的案例中，该公司提供CloudeOS和CloudVision软件，可实现私有或公共云之间的网络连接和管理功能。
Gourlay表示，通过CloudeOS客户可以使用与所有网络抽象的一致运行模型一起运行多个公共云，为所有网络抽象和流程进行操作，以运行其现有的数据中心和校园网络。“与CloudVision耦合的CloudeOS遥测提供了时序存储和对客户多罩网络的网络状态的分析，”他说。“这让客户返回并检查原因和如何发生问题并降低返回操作时间，同时启用初始故障检测的快速根本原因分析。”
在成本方面，单独的Arista提供-CloudeS-CloudeOS边缘 - 在网络边缘支持动态路径选择，让客户分配应用程序的路径以降低数据传输/同步成本。Gourlay表示，通过跨数据中心，校园和多个公共云跨越数据中心，校园和多个公共云进行统一的EOS和CloudVision部署，而不会支持和管理他们的多云策略而不加倍或增加他们的团队规模或影响他们的预算计划。
故障排除
在管理MultiCloud环境中，另一个问题客户面临的脸部是区分应用程序性能问题与网络问题。
“在思科云和计算产品组的副总裁兼总经理Kaustubh Das说：”我们从客户听到的感觉是一种控制权。““难以确定地预测新配置云服务对网络的影响。”
Cisco offers a number of packages targeting the issue, such as its cloud-based Intersight management platform. In addition, Cisco’s AppDynamics application management package and recently acquired ThousandEyes technology, which offers a cloud-based software package that analyzes local and wide-area networks and internet performance. The package is designed to provide broad visibility and let customers pinpoint cloud and non-cloud problems with applications and the network.
Earlier this year Cisco integrated its AppDynamics enterprise application information with Cisco Intersight Workload Optimizer. With it, customers can manage a variety of infrastructure components such as servers, configuration and policy management as well as telemetry and analytics. The idea, Das said, is to let application and infrastructure teams see a shared view of infrastructure dependencies that affect application performance, user experience, and business impact—all from one location.
“IT和Devops团队可以一起使用共享词汇，以确定应用程序劣化的根本原因，积极地防止实时，设置策略中的问题，并自动响应在预定或在云中解决应用问题。域名，“DAS说。
According to Arista’s Gourlay most customers find it extremely difficult to troubleshoot a network issue in the public cloud due to a lack of information and visibility, especially when troubleshooting requires packet-level observability.
“我希望我能够告诉你，它的应用程序和网络伙伴共同努力，使多个罩更有效地制作，但目前不是我们所在的，”Extreme Networks的Nabil Bukhari，CTO和首席产品官员说。
AI / ml的帮助
Bukhari说，将来自物理对准部署和多个云的数据从物理上的部署和多个云进行挑战并将分析应用于挑战，而是一个挑战，这是管理这种环境的关键。
Extreme提供了ExtremeCloud IQ包，提供机器学习和AI驱动的云管理平台，可简化网络基础架构设备的船上，配置，监视，管理，故障排除，警报和报告。
VMware’s Ritthaler said the company’s vRealize Cloud Management includes capabilities to visualize the entire network, both virtual and physical. It uses machine learning to build network and application boundaries and can perform full path analysis across VMs, containers, and across hybrid and multiple public clouds. This offers an easy way to troubleshoot VM-to-VM connectivity across multiple clouds.
“这network topology map also helps optimize network performance with proactive alerting and anomaly detection for firewall misconfigurations, spikes, capacity constraints, etc. All these capabilities are presented in a single network map, regardless of whether the network is virtual or physical or both, and in the context of the application boundaries to make sure the networking, security, infrastructure, and application teams are speaking the same language,” Ritthaler said.
IBM的Frank表示Big Blue通过其应用程序性能管理（APM）包来解决这个问题，它可以帮助客户区分应用程序性能问题和网络问题，跨上所的内部云和混合工作负载 - 所有仪表板。
“在一个多罩环境中，一个同样适用于本地和多个云的解决方案是获得完全可见度和消除孤岛的关键，”弗兰克说。“IBM APM解决方案测量应用程序可用性和性能，自动执行操作以快速从应用程序或网络性能问题中恢复，并提供可见性和工具来诊断和修复问题，然后在其应用环境中影响运营或最终用户的经验。”
内部知识
There are other challenges to managing a multicloud world, observers say.
一个大一个是拥有员工与技能集来管理多思客产品，极端的布哈里说。“数据格式和API是不同的，您需要一个可以了解所有这些东西的团队。”
IDC表示，引入集装箱，微服务和Kubernetes的造成了进一步的复杂性。
“从任何多种式管理组合中获取完全利益，要求组织进行权衡和战略投资选择。在快速移动的技术环境中，难以充分预测新工艺，方法和工具的影响，“IDC陈述。