How to plan to build a large set of Citrix Desktop virtualization architectures-Part2

Earlier, I described how to plan the set infrastructure for a large Citrix desktop virtualization environment, and then we'll show you how to plan the content of the virtual machine hosting section.

After you have completed the architecture of your infrastructure, you are planning for 3000 windows virtual machines. There are more things to consider than the infrastructure part of this piece.

In this case, the blade server is used, all the half knives are used, so we can deploy 16 half knives in a knife case.

650) this.width=650; "src=" Http://s1.51cto.com/wyfs02/M02/76/8E/wKioL1ZWeRTTY-UiAADvverFXj8011.png "title=" hp.png "alt=" Wkiol1zwertty-uiaadvverfxj8011.png "/>

so each half-knife server is installed XenServer Virtualization platform, a total of up to XenServer server, and add it to the same resource pool, just meet the Citrix high limit set up a resource pool standards. Enable HA at the same time to increase the high availability of the entire resource pool.

650) this.width=650; "src=" Http://s2.51cto.com/wyfs02/M00/76/8F/wKiom1ZWePTx8MNXAAD4ipXOUM4448.png "title=" ha.png "alt=" Wkiom1zweptx8mnxaad4ipxoum4448.png "/>

The CPU specifications used in the project are Intel E5-2650v2, which we can see through Intel's website as follows:

650) this.width=650; "src=" Http://s3.51cto.com/wyfs02/M01/76/8F/wKiom1ZWeR3SpcjpAAAtKvqklDg969.png "title=" Cpu.png "alt=" Wkiom1zwer3spcjpaaatkvqkldg969.png "/>

So here's how to plan the density and other performance of each xenserver half-knife server. From the specifications we can see that this CPU is 8 cores, so a half-knife server can have 16 physical cores.

650) this.width=650; "src=" Http://s1.51cto.com/wyfs02/M02/76/8F/wKiom1ZWeV3DeSluAABW1JAfpNg002.png "title=" Vcpu.png "alt=" Wkiom1zwev3desluaabw1jafpng002.png "/>

through citrix The official data shows that each medium load is Socket can support 10 Windows 7 socket can support 5-7 socket support 6 Windows 7 Vm

and to ensure stability, we calculate The XenServer itself needs to consume 2-4 sockets, which areavailable as a large environment . Therefore, each XenServer Half-knife server can support 12*6=72 VMs, rounding up the whole .

So in this case, a knife box with a half-knife XenServer server, so you can support the 70*16=1120 Windows 7 virtual machine at the same time .

Note: At the same time, we also need to pay attention to an important CPU indicator is the frequency, the past history of experience tells us that because the CPU frequency is too low to reflect the user program running slower than the situation occurs occasionally. Therefore, the CPU frequency of the plan should preferably be more than 2.5GHZ, such as in some special scenarios, such as, research and development, design and so the frequency of the proposal to further improve.

at the same time, this example uses PVS Pool Desktop, according to requirements need to have 10% redundancy, so 1000+100 a desktop can be fully supported by a knife box.

but as mentioned earlier, The PVS server itself is not placed on the infrastructure server, so in order to manage the deployment of 4 PVS servers in each box , it is the one that is not deployed in Windows 7 virtual machines.

So in a box, there will be 1100 windows 7 virtual machine + 4 windows Server/pvs to support, that is guaranteed high availability, and stability.

This example requires a desktop, so it is 3 full-equipped knife box to support.

It is important to note that, as a result of a knife box PVs should only support Windows 7 virtual machines in the box , so the 4 PVS servers in each box form a PVs Site .

650) this.width=650; "src=" Http://s2.51cto.com/wyfs02/M01/76/8F/wKiom1ZWeavCK1WTAAAyp74_oK8544.png "title=" PVS Site.png "alt=" Wkiom1zweavck1wtaaayp74_ok8544.png "/>

Note:up to 1 PVs stations can deploy up to 4 PVs servers

During the deployment process, the 4 PVS virtualization servers are distributed across 4 XenServer servers in the same resource pool to ensure high availability.

and as The Write Cache disk for each virtual machine on the PVS pool desktop is not placed on the shared storage, but is placed directly on the corresponding physical server.

650) this.width=650; "src=" Http://s3.51cto.com/wyfs02/M00/76/8F/wKiom1ZWeeuj0hCVAABZbjBWAQU520.png "title=" Bdm.png "alt=" Wkiom1zweeuj0hcvaabzbjbwaqu520.png "/>

It is estimated that the user will be questioned, then this time if the virtual machine or the corresponding physical server failure, it is not impossible to do Vmontion/xenmotion , huh?

in fact, because we are planning here is the pool desktop, so even because of physical machine failure, such as a power outage caused by virtual unavailability, the user can completely re-login to get a new desktop, because we are planning to 1100 a desktop, assigned to the Users, the desktop is a backup. Therefore, it is based on this to achieve high availability of the entire desktop scheduling.

At the same time, the user's personal data and other configuration information through shared storage to ensure high availability. So after the user has re-acquired a new desktop, they can continue the unfinished work.

This article is from the "Citrix in the Cloud" blog, be sure to keep this source http://kaiqian.blog.51cto.com/236001/1717010

How to plan to build a large set of Citrix Desktop virtualization architectures-Part2