Shulou(Shulou.com)05/31 Report--

What is vmare virtualization solution, many novices are not very clear about this, in order to help you solve this problem, the following small series will explain in detail for everyone, there are people who need this to learn, I hope you can gain something.

Recently joined a company! Mainly doing vmare virtualization, although not very close to openstack, but now a lot of openstack will be compatible with vmare program, how will I wait and see, today writing this is also a summary.

Virtualization, cloud computing, big data and smart cities are hot spots in the near future and in the future. Now there are many virtualization products, many companies doing virtualization, and even some application software is approaching "virtualization." When and how do we choose virtualization products for our users, or for our information center managers, in a crowded market?

1 Applications not suitable for virtualization

Virtualization is not omnipotent, and not all applications can be virtualized. We should understand that virtualization is to integrate applications and share resources (including CPU, memory, storage, network), virtualization is to "fight ten"(or even more), and one server after virtualization is used as multiple servers.

Simply put, if an application (or software, or application) is "laborious" to process on a single server, it will be slower to migrate these applications to virtual machines for virtualization, which is undoubtedly not applicable in this case.

Generally speaking, for most graphics, images, Media Processing Service workstations, database servers with high resource consumption, scientific computing applications, these are not suitable for virtualization.

In an enterprise's overall application, although individual applications (such as database servers that require high performance) cannot be virtualized, most other applications can be virtualized. For example, in an application such as that shown in Figure 1, in an application with Load Balancer, the front-end server may run in a virtual machine, while the back-end database runs in a physical machine.

Figure 1 Application with Load Balancer

2 Server consolidation and consolidation can use virtualization

When there are multiple physical servers in an organization, and each server runs only a limited number of applications, most of these servers can be virtualized. In our experience, under current conditions, it is possible to virtualize the original physical servers in the organization on a scale of 1:10 to 1:15 (or even higher). For example, for most government information centers, there are a large number of servers, each of which may be a website or application server for a certain department, and these servers can be virtualized. For a data center with 10 to 20 cabinets and 100 to 200 physical servers, virtualization technology is used, and 8 to 15 physical servers are used instead of the original 100 to 200 physical servers. After virtualization, only 2 cabinets are needed. Moreover, for most organizations, the use of virtualized physical servers does not necessarily require re-procurement, only the selection of servers purchased in the past year or two from existing data centers, and expansion of memory and storage.

For new data centers, upgrades or retrofits, virtualization technologies should be implemented in a planned manner.

Whether upgrading, transforming a computer room to implement virtualization, or building a new data center to implement virtual machines, VMware vSphere is preferred for server virtualization products when selecting products.

3 From physical machine to virtual machine

For server virtualization products, VMware vSphere 5.5 is currently the best choice. While I personally don't advocate P2V (migrating physical machines to virtual machines) during virtualization implementation, using VMware's P2V tool vCenter Converter to migrate physical machines to virtual machines is basically 100% successful. In the process of implementing virtualization, when migrating the original physical server to the virtual server, I personally prefer not to use P2V tools for migration, but to create a new virtual machine in VMware ESXi and then migrate the data to the new virtual machine instead of the original physical machine, so that compatibility will be better. The following is an example of migrating a physical server. The main process is as follows:

(1) Log in to the physical machine to be migrated and record the current hardware information, such as the number and frequency of CPUs, CPU utilization, memory size, used memory, hard disk size, number of partitions, size of each partition, and space used by each partition. Based on these numbers, calculate the number of CPUs, memory size, and hard disk space required by the target virtual machine. In general, the target VM should have CPU utilization below 30%, memory below 50%, and hard disk space below 50%.

(2) Calculate the parameters of the corresponding virtual machine according to the statistical results of the previous step. For example, if the source physical machine has 2 CPUs, CPU frequency is 3.0Ghz, and current CPU utilization is 10%, this calculation requires 2×3.0Ghz×10%<$=2.0hz, while the new ESXi Server CPU frequency is 2.0Ghz, then allocating 1 vCPU can meet the demand. If the source physical machine has 8GB of memory and uses 2 GB, allocate 4 GB of memory for the target virtual machine. If the source physical machine has 3 partitions, each using 40GB, 60GB, and 80GB, then the target virtual machine can be allocated a hard disk of (40+60+80) ×2=360GB. After statistics, virtual machines are created according to the calculation results.

(3) Install the operating system in the virtual machine. The installed operating system is preferably consistent with the source physical machine. During the installation of the operating system, divide the number of partitions by referring to the configuration of the source physical machine. The partition size is twice the original space used by each partition, but the system partition needs at least 40~80GB of space.

If the operating system is Windows Server 2003, the system partition can be kept at 40~60GB, and if it is Windows Server 2008 and later systems, it needs 60~100GB.

(3) After installing the virtual machine operating system, install the application program in the source physical machine against the source physical machine. After installation, backup (copy) data from the source physical machine to the new virtual machine via media such as a network or removable hard disk. After data recovery, configure the application.

(4) Shut down the source physical machine, modify the IP address of the target virtual machine, and provide services to the outside instead of the source host.

(5) Wait for the business to run for a period of time, usually 7 to 15 days. Normal operation, the source physical machine can be used for other purposes.

4 Traditional hot standby can be replaced by virtualization

For the servers of "critical applications" used by hospitals, pharmaceutical companies and financial departments of units, in order to improve the reliability of the system, most of them are configured with "dual hot standby," that is, two servers with the same configuration + shared storage. Operating systems and database applications are installed on the two servers, and the background database is stored on the storage. In this traditional dual-server hot-standby system, the server waste rate is the highest, because at the same time only one server provides services to the outside world, although the other server is also powered on, but the applications installed on this server are in the "standby" state, only when the other server has problems, this server can take over the service and provide services to the outside world. In most dual-computer hot-standby systems, each dual-computer hot-standby system only runs one application. If there are multiple applications, there need to be multiple "dual-computer hot-standby" systems. It can be said that for a unit, the more applications of dual hot standby deployed, the more serious the waste. In addition, when configuring a dual-computer hot-standby system, the hardware selection of servers and storage also far exceeds the requirements of the deployed software. For example, an application system in a hospital is configured with a dual-server hot standby server. Each server is configured with 64GB of memory, 2 Intel E5-2650 CPUs, and 8TB of storage. However, the actual memory of this application during the peak period is only 21GB, and the CPU utilization rate is below 9%. However, the data volume of this system after one year of operation is about 300GB. Traditional dual hot standby is shown in Figure 2.

Figure 2 Dual hot standby

For these dual hot standby systems, VMware vSphere virtualization software can be used to configure HA. It can be said that after HA is enabled, every virtual machine in the cluster (corresponding to each application) is "highly available." If it is said that configuring a set of dual hot standby requires 2 servers and 1 storage, multiple "high availability" applications can be configured using the same hardware.

In general, when setting up vSphere clusters, we generally design a minimum of 3 ESXi hosts (4 network cards recommended for each host), 1 storage, and 1 vCenter Server (placed in an ESXi virtual machine). The virtual machines running on such virtualization clusters and the applications running in the virtual machines are all highly available, as shown in Figure 3.

Figure 3 vSphere High Availability Cluster

The only difference now is that in traditional dual-host hot standby, the time for a host to fail and be taken over by another host is very small, while with vSphere clustering, when a host fails, the virtual machine originally running on this host needs a "restart" time when other hosts take over and provide services to the outside world. Of course, for more than 90% of enterprises, this "restart" time can be tolerated, generally within 1~3 minutes.

If you need a virtual machine with "zero interruption" like dual hot standby, it is not impossible. FT (Fault Tolerance) in vSphere provides this functionality. When FT is enabled for a VM, the same VM runs on two different hosts, one of which is a "primary" VM and the other is a "read-only" VM. When a problem occurs with the host where the "primary" VM resides, the "secondary" VM becomes a "primary" VM to provide services to the outside world. This process can achieve "zero switching time." However, in the current version of vSphere 5.x, virtual machines with fault tolerance enabled can only be configured with one CPU, and it is estimated that this issue will be resolved in the new version of vSphere.

In enterprises with "dual standby", it is relatively easy to implement virtualization, because the standby computer in the enterprise can be the virtualization host (another unused computer in the dual standby system can be installed VMware ESXi at the beginning of project implementation).

5 Traditional Desktop Upgrades Can Use Virtualization

There are a large number of computers running Windows XP in the current enterprise, and these computers need to be upgraded to Windows 7 or Windows 8.1 gradually as Microsoft stops supporting Windows XP operating system. But these computers running Windows XP have low configuration, and upgrading the operating system requires upgrading or even replacing hardware. The traditional way is to replace the new host to adapt to the new operating system. And now with VMware View virtual desktops, there are even more options. Enterprises can solve this problem by configuring VMware View desktops. Without replacing the workstation, the workstation still runs Windows XP operating system, but when users use it, they use View Client to log in to Windows 7 or Windows 8.1 virtual desktop, which is equivalent to running Windows 7 using the user's original hardware facilities. The user's data is still stored on the local hard disk running Windows XP, and if there is a print job, it is still printed on the user's current printer. If there are other peripherals, such as USB cameras and other peripherals, they can be mapped to VMware View desktops for use, the same as using local computers.

After using the View desktop for a while, you can replace the original PC host with a thin client that supports VMware View to achieve energy savings (traditional PC host power consumption is about 130W, while the new thin client power is less than 20W or even only a few W).

6 Storage options in virtualization projects

SAN is preferred over iSCSI storage when virtualization is implemented in the enterprise and FC or SAS direct-attached based shared storage is preferred. Because some of vSphere's advanced features, such as vMotion, HA, DRS, rely on shared storage. Bad storage performance and server-to-storage connection speeds affect virtual machine performance.

Starting with vSphere 5.5, new options are available. Enterprises can also consider using VMware VSAN technology to virtualize server-native hard disks as storage area networks for VMware virtual machines and View virtual desktops. Currently, VSAN requires at least three vSphere hosts with ESXi 5.5 and vCenter Server 5.5 installed, each with an SSD (for data caching) and a local hard drive. VSAN is not recommended as a replacement for SAN at this time, but it can be used as a new storage tier, such as storage space for VDPA backup devices.

When another test site of VSAN is not recommended, if shared storage is used, DRS and DPM functions can be used to centralize virtual machines into several hosts, and idle hosts can be placed in "standby state" to save energy. However, if VSAN is used, this effect may be greatly reduced.

7 vSphere Virtualization Product Choices

When we plan and implement virtualized data centers, we recommend a minimum of three hosts and two storages, as shown in Figure 4. 4 network cards, at least 2 CPUs and 64~128GB memory are recommended for each host. Redundant connections are required between storage, network switches and servers. When selecting 2 storages, if both are FC or SAS storages, you can store VMs on two storages separately and backup on the other storage; if one is FC or SAS storage and one is iSCSI storage, store VMs on FC or SAS storage and backup on the other storage.

Figure 4 Schematic diagram of device connectivity in vSphere datacenter

When using vSphere, configure HA for virtualized data centers and enable DRS, DPM features. In most data centers, when the load is heavy, every host in the cluster will be up and running; when the load is light, DRS will migrate virtual machines and concentrate them on two hosts, while DPM will put the temporarily unused hosts into "standby" state, which will reduce energy consumption. When the load becomes heavy, standby hosts are turned on in turn, and virtual machines are adjusted between hosts.

After vCenter Server and ESXi are deployed, it is best to deploy VDP (Sphere Data Protection) or VDPA (vSphere Data Protection Advanced) backup appliances to back up virtual machines with important applications in the data, and back up all virtual machines if the backup space meets the needs. When selecting a backup location, select a backup device that is not in the same location as the virtual machine. "For example, if the virtual machine is running on FC or SAS storage, you can select another storage for the backup location." When there is no extra storage, you can back up your data on a server's local hard disk space.

As we implement virtualization projects, new storage is purchased. After the virtualization project is implemented, the original storage of the data center can be used for backup use. If the data center doesn't have extra storage, there are tradeoffs between using the saved servers for network storage or using shared folders provided by the servers 'local hard drives for storage locations.

When there are a large number of virtual machines and applications in the datacenter, you need to select vCenter Operations Manager to dynamically monitor the vSphere datacenter. vCenter Operations Manager collects performance data from every object at every level of the virtual environment, from individual virtual machines and disk drives to entire clusters and data centers, stores and analyzes that data, and uses that analysis to provide real-time information about problems or potential problems anywhere in the virtual environment.

8 Conclusion

Virtualization is not omnipotent, but it is impossible without virtualization. With the development of enterprise informatization, the increase of various applications, and the change of concept of unit leaders and information center managers, virtualization as a basic application, its development will be better and better. Organizations are virtualizing everything from data centers to clouds to mobile devices with VMware software, simplifying enterprise IT operations and making businesses more agile and efficient.

VMware pioneered virtualization and policy-driven automation technologies to simplify IT complexity across the data center. VMware delivers value to more than 55,000 partners and 500,000 customers through virtualization software, professional services and a robust partner ecosystem, enabling application interoperability and providing customers with greater choice.

Did reading the above help you? If you still want to have further understanding of related knowledge or read more related articles, please pay attention to the industry information channel, thank you for your support.

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.