fifth or sixth grade in about to begin teacher points deducted when units not included (inches, pounds, meters, etc.) with numerical answers to word problems in math tests. During a recent evaluation of Citrix XenDesktop on the NetApp FAS8060 All-flash storage platform, I confess that I was guilty of forgetting to add units The majority of XenDesktop test benches.
I had previously used memory configured shelves of hard disk drives (HDDs) or hybrid storage solutions with hard drives and a small number of solid state drives (SSDs). Frankly, I'd get used to read latencies of a few milliseconds and write. When I saw latencies 450-1700 way, I thought was something wrong, and I had made a mistake, the test environment at the facility. It took a moment to realize that these latencies values in microseconds and not milliseconds. This "aha" moment lifts the low latencies of NetApp All flash FAS solution and how they can help desktop virtualization provisioning and boot times for XenDesktop to accelerate.
The objective of this test exercise was to evaluate the NetApp All flash FAS storage platform with an I / O-intensive pure virtual desktops (VDI) workload. In particular, I wanted to characterize I / O performance for XenDesktop deployment and boot storm scenarios, the results of using Machine Creation Services (MCS) and Citrix Provisioning Services to compare (PVS) for providing a 2,000-seat VDI Farm. I wanted to examine the effects of storage efficiency features such as deduplication. I have the test configuration, methodology and test results in a short white paper (here) documented that this blog summarizes.
The audit
In addition to the FAS8060 storage platform from NetApp, the testbed included (Figure 1) two network switches and 16 rack servers (14 for VDI desktops and 2 for infrastructure services). Software components include Citrix XenDesktop 7.6, Citrix Provisioning Services (PVS), Microsoft® System Center Virtual Machine Manager 2012 R2 (SCVMM) and Microsoft® Windows® Server 2012 R2 with Hyper-V®. The NetApp system presents multiple iSCSI datastore on the Microsoft Hyper-V hosts for desktop storage needs.
I led two separate test phases, one with MCS and PVS, provisioning 00 persistent VDI desktops in both cases. collected At each stage I I / O metrics during desktop provisioning process and in the same boat storm scenario where all 2,000 seats were started simultaneously.
provisioning and boot storm results
Further, as shown in the table below, FAS8060 system accelerates NetApp strongly MCS provisioning and boot times. MCS Provisioning took less than 41 minutes to boot providing all 2,000 desktops and 16 minutes the desktops in a boot storm scenario. PVS took 88 minutes for the provision and 25 minutes to boot. Since PVS also write cache drives need to be copied during the deployment process, it makes sense that the provision takes longer. How PVS in expected significantly lower IOPS, during the initial deployment process and 11K as compared to 68K during the Boot Storm Test 227 compared to 41K for MCS average.
The results show that MCS is an appropriate deployment option for this size workload, since the storage system can provide sufficient IOPS. Because the NetApp FAS8060 All flash functions such as low latency, it is able to provide the necessary storage system performance for both MCS and PVS provisioning and boot operations.
efficiency analysis
Each desktop virtualization project, the cost is always a concern, and storage can often the most expensive component in the overall solution be. The NetApp FAS8060 platform and clustered Data ONTAP 8.3 operating system make more efficiencies possible for the aid to reduce the amount of memory, which can save costs.
As shown in the figure below, the NetApp storage solution allows a considerable reduction in capacity requirements for both MCS and PVS by the use of thin provisioning and always on dedupe technology. Additional memory savings would be always achieved with the additional NetApp space efficiency technologies like inline zero-elimination, sis clones and inline compression. Storage capacity for the base operating system of 40GB reduces both MCS and PVS to 24GB. would Without thin provisioning and deduplication applied MCS need about 2.8TB differencing disk for every 2,000 desktop support. With thin provisioning and deduplication only about 1.2 TB is needed, a reduction of 60%. Since PVS provisioning would require storage for write cache drives 10TB without the use of these efficiency features are needed. With thin provisioning and deduplication PVS requires only 1.2 TB-a 88% savings.
Conclusion
These tests exercise shows how the NetApp All flash FAS8060 system can provide low latency and satisfy IOPS for requires either MCS or PVS deployment of 2,000 persistent VDI desktops. Whether MCS or PVS is used NetApp always-on storage efficiency technologies reduce storage needs. (A new feature in Data ONTAP 8.3, Advanced disk partitioning also increases the capacity for data, as it eliminates the need for a dedicated root aggregate.)
Citrix and NetApp engineers together to help in many reference architectures and test scenarios customers with the design appealing, highly available and cost-effective desktop virtualization solutions. Many of these solutions are documented-inclusive offer whitepaper on this test exercise on best practices and policies, build a successful solution architecture to help. As always, it is recommended that you implement to collect metrics with typical workloads is a proof-of-concept to help suitably an insert size.
For All flash FAS8060 system and the results of the read more about the NetApp see XenDesktop tests, the complete test report White Paper and NetApp recently published reference architecture TR-4243
- , Rob Briggs, principal solutions architect with Citrix Worldwide Alliances
0 Komentar