Microsoft Cloud Iaas&paas Big Comparison

1, for IaaS, the objects that users need to manage are:

-O/S, operating system

-middleware, middleware

-runtime, run-time

-There's data and application.

For IaaS, customers are like virtual machines in the cloud, selecting their own operating system (Windows or Linux) on the Azure platform, installing the required software on the operating system, and then managing the operating system and the above content.

What is the advantage of 2.IaaS?

　　The benefits of IaaS are first and foremost similar to the traditional IDC operations (note is similar but not the same, because public clouds are based on virtualization technologies, such as cloud network virtualization is different from traditional IT operations), traditional IT operations are easier to accept and quickly get started. In general, when I do the project implementation, the cloud virtual network, storage accounts and virtual machines are created, directly to the Windows Remote Desktop or the Linux SSH connection string to the customer it party, you can quickly install the required software.

The disadvantage of 3.IaaS?

　　The disadvantage of IaaS from my personal understanding is divided into several:

　　(1) It still needs to spend time updating and maintaining the server

Public cloud services emphasize self service, and for the IaaS level, while enterprise users don't need to invest a huge amount of money to build and maintain their own data centers, they still need to spend their time updating and maintaining the operating system, such as it or the need to upgrade and patch the operating system, for maintenance. The use of IaaS saves hardware procurement costs, but does not address software maintenance costs.

(2) Users need to configure their own high-availability groups

Azure provides update domains and fault domains to enable high availability of servers. On the IaaS platform, users need to manually configure high-availability groups and configure multiple virtual machines with the same functionality in the same high-availability group. Otherwise, the virtual machine cannot be highly available.

(3) In order to achieve high redundancy, we need to change the architecture

At the beginning of the launch, public cloud services were calculated in parallel with multiple virtual machines to provide a high level of redundancy to solve the problem of server single point of failure. We know that any hardware server will have a hardware failure. On the Azure platform, we recommend that users typically use 2 or more 2 services in the same high-availability group to achieve a 99.95% SLA. But the problem comes, and I find a lot of customers using:

A. A single Web server as the front end, and the session is saved on the Web server. Once this Web server goes down, all sessions are lost and the system does not function properly. If we consider a 99.95% SLA, we need to use 2 Web servers and detach the session, using the cache service provided by the cloud provider or implementing the master-slave logic of the cache in the virtual machine itself.

B. The unit file server holds files. This file server also generates a single point of failure. If you need to commit files to multiple nodes at the same time, you need to use rsync on the Windows platform, such as DFS (Distributed file system) or on a Linux platform. Implementation files are committed and synchronized on multiple nodes.

C. Single SQL Server database server. The database server has a single point of failure. We can use SQL Server always on to make the database highly available.

(4) Limited scale-out capability

　　If you are using Azure Auto scaling, you will find that at the IaaS level, if you need to implement Auto scaling for 100 virtual machines, you need to manually pre-create 100 VMS and configure the same application in each virtual machine. For IaaS, Azure Auto Scaling does not automatically scale to any size and requires IT managers to make presets. The main reason for this is the data synchronization problem between the IaaS VMs, and Azure has no way to deploy the application scale-out to 100 VMS, because the underlying operating system files and application files for each virtual machine need to be managed by the user themselves. Although we can quickly deploy multiple virtual machines through virtual machine mirroring, it involves IP rules, data file synchronization, and so on, all of which require the user to manage. So for IaaS, scale-out is limited.

We talked about IaaS, and then we talk about the PAAs platform. As shown in the top figure:

For PAAs platforms, the objects that users need to manage are:

-application, Application

-data, data files

　　All the underlying os,middleware and runtime are maintained by Windows Azure. be aware, however, that the operating system provided by Azure PAAs is actually windows. Linux operating systems cannot be used at the PAAs layer. If your app must run on a Linux platform, use Azure IaaS.

What are the advantages of the 4.Azure PAAs platform?

(1) No need to maintain operating system

The advantage of the Azure PAAs platform for the enterprise is the value of the service, not the cost of implementing and maintaining the software system.

Azure PaaS Cloud Service defines osfamily and osversion to automatically configure the operating system version and update content, without requiring users to manually upgrade the operating system. If the customer needs Azure to automatically upgrade the latest operating system, it will osversion= "*"; if the customer needs to fix a version of an operating system, you can set osversion= "Wa-guest-os-x. Xx_20xxxx-xx ", set a fixed version.

　(2) Automatically configure high availability groups

If a user's application requires load balancing, the Azure PAAs platform automatically deploys the application to multiple compute nodes and automatically sets the fault and update domains. The user does not need to be configured manually.

(3) Automatic scale-out

　　The Azure PaaS platform packages the user's code into CSCFG and cspkg. Cscfg is a configuration file that describes the number of nodes and configuration information for the PAAs. CSPKG is the project packaging file (package). Azure PAAs can automatically implement scale-out by defining compute nodes through CSCFG and then publishing cspkg to several nodes. Unlike IaaS, it admin needs to pre-create the virtual machine.

　　(4) Highly redundant and scalable architecture

　　Azure PAAs platform architecture allows for high redundancy and scalability of applications.

By saving the session to an external store, such as a SQL Server database, you can implement high redundancy for the Web server.

By saving the file to Azure storage, you can achieve a high availability of 99.9% of the file. In addition, the files are not uploaded to the compute nodes, which resolves the problem of file synchronization and consistency.

Disadvantages of the 5.Azure PAAs platform

(1) need to help developers

　For many traditional it deployments, application publishing needs to be implemented with it. With Azure PaaS, the deployment of your app needs to be developed with the help of developers. It's a challenge for it.

　　(2) Non-persistent VMS

Azure PAAs is a non-persistent VM, and any software installed on the PAAs platform is non-persistent. Once the PAAs VM is reset, the previously installed software will be lost.

For large applications, it is common to install other third-party software (such as exporting Excel, exporting PDFs, other third-party plugins, and so on).

For PAAs platforms, the steps to install the software need to be placed in the startup task of Servicedefinition.csdef for silent installation (-s)

(3) Need to modify the code

　　The Azure PAAs platform needs to modify some code, such as the session is placed on external storage, local file reads and writes to the Azure storage. I'll cover the steps of IaaS migration to the PAAs platform in the following chapters.

　　Summarize:

　　

Azure IaaS is better for:

-Migration of old applications

-Easy migration, but more complex management

-Enable migration of applications with it

-First sweet after bitter

Azure PAAs is better for:

-Development of new applications

-Migration difficulty is complex, but management is relatively simple

-Need developer, code modification and development based on PAAs

-First Bitter sweet

Microsoft Cloud Iaas&paas Big Comparison