posted

2 Comments

In 2017, we kicked off Getting More Out of vSAN, an ongoing technical webcast series led by vSAN experts and designed to support you, our customers, with deployment and operation of vSAN. This webcast series is unique because of the interactive, live Q&A session at the end, where you can voice your concerns, provide feedback, and get your questions answered live by the vSAN product team!

After each session airs, we will share a high-level recap including the top five questions and answers, right here on Virtual Blocks. If you have 60 minutes to spare, we also highly recommend that you watch the full, recorded session by registering here.

 

Session 1: vSAN Architecture & What is New in 6.5

In this session, Jeff Hunter, Staff Technical Marketing Architect and Vijay Ramachandran, Senior Director of Product Management, provide a thorough overview of vSAN’s architecture, commonly used terms, and the benefits of its unique in-kernel design.

They discuss the key advantages of the Storage Policy Based Management (SPBM), which allows for VMs to reside on the same data store but have different characteristics assigned to them through policies. Changing the policies for a VM is online and does not require a reboot or migration.

The webinar also covers fault domains, the differences between implicit and explicit fault domains, and dispels myths surrounding Data Locality as it relates to vSAN i.e., the idea that traversing the network introduces meaningful latency.

Next, they discuss vSAN stretched cluster and 2-node configurations and use cases. Stretched clusters enables customers to have Active-Active sites. vSAN also comes with a unique per-VM licensing, which minimizes the cost for small deployments – so if you are running less than 25 virtual machines in a Remote Office Branch Office location, do look into vSAN ROBO licensing.

The webcast concludes with insight into the new features of vSAN 6.5 including 2-node direct-connect, automation with PowerCLI, and iSCSI access for physical servers.

 

Top Five Questions and Answers from Session 1

Q1. Nutanix claims data locality matters. What is wrong with this argument?

A: Data locality initially sounds like a good idea. Close affinity of data to a virtual machine would seem to give the best performance. However, there are two key reasons why this argument falls apart: 1) data locality becomes a huge problem during a vMotion since data becomes not local and negatively impacts performance of all VMs, 2) data locality is a problem during node failures since all data must be moved.

vSAN introduces a more elegant distributed data algorithm that provides excellent performance in optimal operations as well as during vMotion and node failure events.  With an embedded in-kernel architecture that leverages intelligent read caching on low latency networks, vSAN delivers better performance than Nutanix without the drawbacks of data locality

Read an in-depth discussion on this with real life examples here

 

Q2. Does vSAN allow for fault domains overlapping each other?

A: That is what we call a "Nested Fault Domain". Stay tuned on this feature. It is on our roadmap and currently planned for 2017. Nested Fault Domain essentially refers to local protection for stretched clusters. It allows you to protect your data within a site and across sites at the same time in a stretched cluster. In other words, RAID-1 across sites and then RAID-1, 5 or 6 within a site. We are looking forward to bringing you this very exciting functionality!

 

Q3. All-Flash was made available in all editions of vSAN. Is this retroactive to vSAN 6.2?

A: Yes, it is retroactive. Customers with vSAN 6.0/ 6.1/6.2 STD can add all-flash support to their licenses by converting their licenses to a “vSAN 6 with All Flash Add-on” license. You can convert your license through the license “downgrade” process in the MyVMware portal. Here is what it should look like:

MyVMware

More information on All-Flash support for vSAN 6.0/6.1/6.2 STANDARD licenses

 

Q4: How can you upgrade from vSAN 6.0 to 6.5? Can you upgrade in place?

A: Yes, you can upgrade vSAN by upgrading vSphere from 6.0 to 6.5. VMware Update Manager is recommended to ensure a smooth upgrade. The vSAN upgrade comprises of three steps

1) Upgrade vCenter Server

2) Upgrade ESXi

3) Upgrade on-disk format

Upgrade vSphere (1. Upgrade vCenter Server 2. Upgrade ESXi)

The two core components of vSphere are VMware ESXi™ and VMware vCenter Server™. You can upgrade the vCenter Server system on a Windows virtual machine or physical server, or upgrade vCenter Server Appliance. More information on upgrading vSphere

Upgrade vSAN (3. Upgrade on-disk format)

This process is online and non-disruptive. If you have adequate capacity in the cluster and have more than three hosts and/or disk-groups, data is not exposed to risk during this operation. However, this process is I/O intensive and can result in reduced performance for production workloads until the process is completed. More information on upgrading on-disk format

For detailed steps on upgrading to vSAN 6.0 and on-disk format, read this blog

 

Q5: Does vSAN support 1 Gb NICs?

A: vSAN supports 1Gb connectivity for hybrid configurations (suited mostly for smaller deployments). However, we recommend 10 Gb for hybrid and require 10 Gb for All Flash. For most enterprise workloads, especially on All Flash, we require a minimum of 10Gb due to the amount of I/O we are able to serve to meet high workload demands.

 

Thanks for reading! Get your own questions answered and grow your vSAN expertise when you register for upcoming and on-demand sessions.