Since we launched the vSAN Assessment last fall, over 300 customers have ran a free, one-week comprehensive analysis of their vSphere environments to see where and how Virtual SAN can save them money and time. Results include information on all aspects of Virtual SAN, ranging from environmental infrastructure analysis to financial estimates of overall storage CAPEX savings.
The tool collects detailed performance statistics on the targeted virtual machines; data that is often times difficult to gather using monitoring tools included with traditional SAN solutions. The reason it’s difficult to assess in traditional non-HCI environments is because of the way storage is presented and consumed in a virtual environment. SAN administrators typically log in to a dedicated interface for the SAN, configure the physical disks in some sort of tier or group, assign a RAID level, and map a LUN to the vSphere hosts running in their environment. At that point, it’s the vSphere Administrator’s turn to format the new LUN in VMFS, and start provisioning and/or migrating virtual machines onto the newly presented volume. The SAN can then monitor performance on the LUN that is presented to the vSphere environment: information such as total IOPS, throughput, and reads/writes is readily available to the storage administrator. The challenge is that most customers have a number of virtual machines running on the LUN in question, and it becomes problematic to figure out which individual VMs are contributing to the overall workload the SAN is servicing. Additionally, any hyper-converged solution has to support both storage and compute workloads, so design and sizing becomes a daunting task.
Enter the VSAN assessment tool, which uses the vSphere hypervisor to track storage performance on a per-VM basis. The solution gathers read/write IOPs, read/write bandwidth(KB/s), and analyzes patterns to estimate how suitable the particular workload is for VSAN. At the conclusion of the assessment each VM receives a score from 1-100, based on compatibility with either a Hybrid or All-Flash VSAN deployment. We can confidently recommend a type of deployment because we have all the statistics we need: read cache size, write buffer size, minimum usable capacity, per-VM profiling with the average number of VMDKs, vCPU, and vMEM per VM. This ensures that the proposed solution has all the horsepower necessary to run a customer’s workload, from both a storage and compute perspective. Make sense? Great. So now you may be wondering, how can I get an assessment, what is the process and what are the requirements, and how long does it take?
If you’d like in-depth information about the underlying architecture used for the assessment, please refer to this blog post by Rawlinson Rivera.
How can my organization get a VSAN Assessment?
The VSAN assessment is a cloud-based, SaaS solution that is delivered by one of VMware’s partners. Corporate resellers, value-added resellers, consulting organizations all have the ability to run an assessment for your organization.
Note: Please specify if you have a preferred partner that you enjoy working with, otherwise once you register we will recommend a local partner to assist with the assessment process.
What happens next? Process and Requirements
Your partner will reach out and set a time to begin the process, and you’ll receive an email inviting you to participate in an assessment and download the collector appliance. The collector appliance is in OVF format, making it very easy to deploy in your existing vSphere environment.
Requirements (full list of requirements list is available in the download package):
- Collector Appliance – 4 vCPUs / 8 GB of RAM / 200GB disk space (will be released at conclusion of the assessment)
- IP address for the appliance – Must be able to route to vCenter and management IPs for all the underlying vSphere hosts being monitored for performance
- vSphere 5.0 update 1 is the minimum supported version
- VIB install on ESX hosts necessitating maintenance mode while VIB installs – To do automatically the customer needs DRS enabled, manual install can be performed if necessary, but still requires maintenance mode
- Service Account with admin privileges to be used during the collector configuration
The overall setup and deployment is fairly straightforward, the most common challenge is with the VIB installation. If your vSphere clusters are configured for DRS with fully-automated vMotion then it’s a snap: you can automatically install the VIB and the collection process will begin. If your environment doesn’t use DRS in an automated fashion, you can manually place each host into maintenance mode and install the VIB, which is more time-consuming but equally as effective. The video at the end of the post will help walk you through the process.
How long does it take?
The recommended run time is 7 days of performance data collection. This time period can be lengthened or shortened depending on your particular needs. The best practice is to make sure it’s running during the busiest week of any particular month, for example, during month-end procedures that generate more of a storage workload.
Brand New Video Walk-Throughs
The team put together a couple videos walking through the assessment process, including registration, downloading, and deployment of collector appliance, VIB install, etc. A few customers received the beta version of these videos and the deployments were seamless, please check them out, they are very helpful.