The new write Cache Technology of PVS reduces the iops demand of virtual desktops by 99%

Over the past years, hardware costs have been a nightmare for VDI projects. It is often to meet the minimum iops required for Windows 7/Windows server operating systems, users have to purchase expensive storage devices, or because of budget constraints, they can only reluctantly; even if you have the courage to purchase desktop virtualization, you can also choose a small scale pilot. In many cases, the project is packaged with hardware for bidding, as a result, virtualization software often only accounts for a small proportion of the overall project budget, and often loses the right to speak. System integrators often recommend high-grade storage devices for various reasons. In fact, even before the release of the new PVS write cache technology, the VDI project only requires the iops and capacity indicators. Even the lowest-level storage can complete the deployment and implementation tasks perfectly if the two conditions are met; the various management functions provided by high-end storage devices are basically useless in the VDI project. Therefore, we recommend that you deploy a project by using DAS.

But now, a new era is coming. Not only does it not require high-end storage, but it has become history, thanks to the new PVS technology "cache in RAM with hard disk overflow ". In the future, do not talk about iops.

If you do not know what "cache in RAM with hard disk overflow" is, let's simply combine this technology. For more detailed analysis documents, refer to my previous blog: PVS to shake storage, one of the series of blog posts: PVS's new write cache technology. "Cache in ramwith hard disk overflow" is a new write cache method introduced by PVS after version 7.1, this write cache method first places the I/O requirement on the non-paged pool memory. When the memory you allocate is full, PVs reads and writes the remaining data to a vdhx file deployed on the local hard disk.

Since the acquisition of argS changed its name to PVS, the write cache was used for the writecache file type in vhd format until the emergence of the "cache in RAM with hard disk overflow" technology, this traditional method is also the cause of the aslr problem (to solve this problem, we developed the New Technology "cache in RAM with hard disk overflow ), this is different from the write cache file format in the vdhx mode of "cachein RAM with hard disk overflow", so you need to know the following points:

• If you think that the PVS technology has entered dusk, this is a big mistake. Please try "cache in ramwith hard disk overflow;

• "Cache in ramwith hard disk overflow" is an epoch technical product, and you do not need to consider iops requirements in future projects. In the past, you had to consider designing 15-25 iops per VM,What you need now is that each VM only needs 0.1 iops;

• Do not think that to implement "cachein RAM with hard disk overflow", you need to increase the memory allocated by each Vm, or allocate a large memory to the VM. In the lab number, even if only MB of memory is allocated to the Windows 7 virtual machine, 2 GB of memory is allocated to the xenapp server to obtain excellent results;

• Are you still worried that I/O is insufficient if the memory overflows to the hard disk? You need to know that when you use the write cache in the previous vhd file format, most random write operations are 4 K or 8 K, the read/write size of this file is the nightmare of the hard disk; the new write cache in vhdx format uses 2 MB file blocks for read/write, the size of the file block is easy for the hard disk, and there is a lot less iops, so even with the same read/write volume, the old and new write cache technologies also have a huge difference in the hard disk pressure;

• Or are you worried about the storage capacity of the hard disk after overflow? I will discuss this issue in a new blog post. In short, the write cache on the hard disk can also be fully configured with a streamlined configuration. Of course, hard disks are no longer worth it now, haha

• Are you worried that the new technology will not be implemented by customers and you will become a mouse? At present, at least three giant customers have already implemented this technology, and from the perspective of field performance, the performance is even better than our test data, because we use the moderate load pressure of loginvsi, in fact, users generally do not have such a high workload; as for the test site, the performance is equally good, and many on-site test projects have confirmed this;

• Do you need a special version of PVS? No. Please download the latest version of PVS software at: http://support.citrix.com/article/ctx140875. directly click the following address:

  Http://support.citrix.com/servlet/KbServlet/download/37672-102-711668/CPVS71003.zip

• If you are still trying to use vsan, you are planning to use an SSD hard disk or find a more cost-effective storage solution to solve the iops problem. Please stop and try this new technology of PVS.

   

This article is from the Citrix virtual world, you have me blog, please be sure to keep this source http://virtualworld.blog.51cto.com/1412963/1544705

The new write Cache Technology of PVS reduces the iops demand of virtual desktops by 99%