Shulou(Shulou.com)06/02 Report--

Like a trend, everyone is talking about the cloud now, even more than the virtualization of earlier years. With the help of major IT manufacturers and Internet enterprises, users are scrambling to build clouds and Shangyun, and even a small-scale enterprise or unit may manage "multiple clouds" at the same time.

However, due to the dispersion of construction and management, it is often difficult to find a person who can clearly and comprehensively explain the characteristics and differences of each cloud, and very often only know its business scope. Whether there is excessive design, waste of resources, who is responsible and other issues may be a lot of confusion, the only certainty is that a lot of management difficulties and manpower investment have been increased.

Let me introduce the definition of cloudy clouds, cloudy nanotubes and their value to clear up some misunderstandings of basic concepts.

How do cloudy clouds define the distinction

IaaS cloud

In fact, IaaS cloud is not rigorous, it just borrows a conceptual analogy to describe a type of cloud platform that focuses on providing infrastructure resources. Typically, there are computer rooms built by enterprise units and traditional IDC hosted data centers, which are defined as IaaS, which is more of a need for commercial promotion or project name.

In the early days, this kind of cloud generally provided only the basic services of computer room facilities, power surround, and hardware equipment, such as delivering to users after the machine was powered on, as well as some daily hardware alarm check and notification services.

With the popularity of virtualization technology, when multiple computing and storage nodes are integrated into a pool, the original pile of X86 machines is upgraded to what people call a "cloud platform", which is as flexible as water and electricity. This is a cloud defined from a technical level.

PaaS cloud

With the business transformation and scale expansion of the industry, users have put forward more demands for management services. For example, in addition to virtual machines, resource services also need database services, middleware services, network services, security services, and even some advanced big data services, micro services, Devops services, etc., in addition to basic hardware status monitoring in management services. It also requires business system monitoring, centralized resource management, automated operation and maintenance operations, process dispatch, reports and portals, and so on.

Traditional IDC found that "IaaS cloud" is no longer competitive, and virtualization alone can no longer satisfy users. With the gradual rise of the Internet, the industry gradually found that a set of "operating system" for cloud characteristics is needed. At this time, the most representative Openstack arises at the historic moment.

Internet leaders and operators have customized their own PaaS service clouds based on it, such as Huawei Cloud, Tencent Cloud and Telecom Cloud. First of all, they work in the public cloud domain, gradually sinking to the private cloud, and even changing to more fashionable names such as data console, application console, AI console and so on. This is a cloud defined by shielding the underlying infrastructure and focusing on differentiated services.

Industry cloud

What we often hear from users, such as government cloud, police cloud and education cloud, are more customized based on the above IaaS and PaaS clouds, configuring different levels of resources and service types according to the business characteristics of industry applications, and more often a cloud defined from the business scope.

Regional cloud

In addition, there are some clouds named after provinces, cities, autonomous regions and counties, or direct unit names, which are likely to be similar to IaaS and PaaS clouds in nature, which will not be discussed here.

How to manage in a cloudy environment

When we roughly distinguish the various cloud platforms, we are soon faced with a series of questions about how to manage them uniformly, including:

Can it be managed uniformly between public and private clouds and between cloud platforms of different vendors?

If you can manage, regardless of the granularity, can you initiate the creation, switching and snapshots of virtual machines on multi-cloud platforms in one stop?

Is it true that as long as you take over the cloud platform, all virtual machine systems on it will be uniformly managed?

To answer this question, we need to split the cloud platform into three parts: the cloud platform itself, the virtual machine system components and the PaaS service components.

Cloud platform

In essence, it is a set of operating system for cloud environment, similar to the upgraded version of Windows and Linux. Most of the products we use are platform products that have been commercially added and packaged by manufacturers, which provide a variety of convenient services, such as viewing resource pool capacity, multi-tenant isolation authorization, and virtual machine additions, deletions and removals. These functions are solidified within the platform and can only be called through the API disclosed by the manufacturer. Otherwise, the actions that can be driven only by basic network access and a small number of open source interfaces (some manufacturers may) are very limited, which is an objective technical limitation.

Virtual machine system component

That is, the operating system, database, middleware and applications in the virtual machine delivered by the platform (strictly speaking, the virtual machine hardware is also a part of the cloud platform), in essence, they are no different from those delivered by the physical machine and virtualization platform, and are basically independent of the underlying platform, and can still be managed in accordance with conventional methods, such as Agent.

PaaS service component

PaaS Cloud provides more "packaged" database services, middleware services, network domain name services, security services, big data analysis services, micro services and so on in the way of service catalog, which is often based on the transformation and encapsulation of open source components, which is different from traditional components, so it cannot be managed in a conventional way and needs specific analysis. It is often implemented indirectly through API, log, network access protocol and other ways.

What is the value of multi-cloud tube?

Suppose we can really manage multiple cloud platforms from the above three dimensions, but what is the value of simply putting things together? We can try to think from the perspective of operation and maintenance scenarios:

1. The need for unified management and allocation of resources, business personnel do not need to care about the underlying resource types, only need to apply for resource allocation by on-demand bill of lading, dynamic scheduling and cost accounting by the unified platform.

two。 Unified monitoring is required, especially for business systems that are distributed on and under the cloud at the same time, or need to pay attention to the monitoring requirements of hardware, cloud platform and system components at the same time. Only in this way can we really start from the business and improve the efficiency and quality of monitoring.

3. It provides the possibility of automation across cloud platforms, registers the drive interfaces or data interfaces of various dimensional resource components on different platforms with the unified platform to form atomic components, and custom scheduling is arranged as a standardized process to meet the requirements of complex tasks across platforms, such as virtual machine migration in theory.

4. In addition, there may also be unified process work order services in a multi-cloud environment, unified data operation visualization, and comprehensive big data analysis.

In the case of multi-cloud unified management, these scenarios may have endless room for imagination, waiting for users, manufacturers, service providers and all parties in the ecology to explore together.

Author: Zeng Jiacheng

Previous articles

How does DevOps achieve efficiency improvement?

Tencent PaaS Influxdb node migration

What about the explosion of archived files after Oracle data migration?

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.