VMworld 2017 US Recap

This year, I was rewarded with a blogger pass for VMworld.

In this blogpost I want to highlight, what was striking to me.

Solutions exchange

Every year I like to walk around the Solution Exchange, talking with vendors and learning about their solutions.
What was surprising to me this year, was the size and the quality of boots. Comparing to other VMworld, the boots where much larger and had more lighting and some of the even LED screens. I know that the development and a spot on the Solution Exchange is not cheap, so clearly these vendors see the importance of being present and want to invest in an event like VMworld.

Sessions

I’ve attended a couple of sessions, as most session are recorded and can be watched online after VMworld. The sessions I’ve attended, these sessions were striking to me.
Most of these session will also be in VMworld Barcelona.

HCIbench: Performance Benchmarking and More

  • Speakers:
    • Dave Morera
    • Chen Wei
  • Twitter: #STO2865QU

This fling is a Hyper Converged benchmarking tool. Although this is a fling, the output from HCIBench can be imported in VMware vCenter 6.5. When imported, you can troubleshoot potential performance issues. The integration with VMware vCenter is very nice, as it combines the output from HCIBench with the performance metrics in VMware vCenter. This gives you a in depth view where a potential problem occurs.

A Deep dive into vSphere 6.5 Core Storage features and functionality

  • Speakers:
    • Cormac Hogan
    • Cody Hosterman
  • Twitter: #SER1143BU

It has been a while since this session was on VMworld, but the room was packed! The speakers gave a indept view about the new features in VMware vSphere 6.5. Wat was surprising to me, was the new virtual NVMe adapter available in vSphere 6.5. This adapter enables you to configure a virtual NVMe adapter for a virtual machine. If the vSphere host has NVMe storage, you can benefit from the NVMe architecture, leveraging a better performance as you don’t have to use SCSI commands.

 Extreme Performance Series: Byte-Addressable Nonvolatile Memory in vSphere

  • Speakers:
    • Qasim Ali
    • Praveen Yedlapalli
  • Twitter: #SER2734BU

This was a tech preview session, and it’s not clear when or if these features come into VMware vSphere. But the speakers showed a custom vSphere 6.6.1 build with custom HPE hardware (2 x Intel Xeon Gold 5117 28-cores with 1GB pMem). They used this setup to test Byte-Addressable Non Volatile memory, aka Persistent Memory, aka pMEM. You can use this pMEM in 3 configurations:

  • As a datastore
  • Connected to a virtual machine
  • Connected to the guest OS

The presenters have done several tests, and as you can imagine, the performance was mind blowing. High IOPS and low latency. Although the capacity of 1 GB we won’t see virtual machines running from pMEM, you can benefit from this was type of storage. When attached to the virtual machine or guest OS, you’re able to host certain type of files (SQL log files for example) on pMEM so that application runs a lot faster, up till 8 times! Do mind, when attached to the guest OS, the application must support pMEM to benefit from it.

 Unconvering ESXTOP

  • Speakers:
    • Jagadeesh Devaraj
    • Ramesh Nataraja
  • Twitter: #SER2905BU

I’ve been using ESXTOP for a long time now, especially when I have to troubleshoot performance issues. This session was to explore ESXTOP and how you can use this tool. What really was interesting for me, was the correlations between the different metrics. Ihe speakers explained very nice how ESXTOP metric correlated. If you’re attending VMworld in Barcelona, I would recommend attending this session. Otherwise you can watch this session online.

Extreme Performance Series: vSphere Compute & Memory Schedulers

  • Speakers:
    • Xunjia Lu
  • Twitter: #SER2343BU

This was a real nice session, deep diving into the vSphere Compute & Memory scheduler. Although it had some overlap with the Unconvering ESXTOP session, this session deep dived into the compute and memory scheduler. The speaker explained in detail how to use the ESXTOP metrics and how the correlate with each other. During the session, the speaker showed a screenshot of ESXTOP CPU %Ready value of more than 25%. He asked the audience, how thought that this value was to high. Most of the attendees (inclusing myself) showed hands. As you probably know, %Ready shouldn’t be higher dan 5%. Than the speaker showed another screenshot where he expanded the virtual machine world show you can see a breakdown of all resources for this virtual machine. It became clear that this virtual machine was configured with 32 vCPU’s and each vCPU had a %Ready value lower than 1%. The sum of all %Ready values of all vCPUs was more than 25%. The speaker taught me a valuable lesson, never take anything for granted.

vSphere 6.5 Host Resource Deep Dive: Part 2

  • Speakers:
    • Frank Denneman
    • Niels Hagoort
  • Twitter: #SER1872BU

This is part 2 of one of the best session at VMworld in 2016. Frank and Niels wrote an excellent book about Host Resource Deep dive for vSphere 6.5 and in this session the picked a few topics to present. I’ve been to a couple of sessions of both speakers, with similar content, but every time I’m surprised about how in-depth they tested and explain CPU, Memory, Storage and Networking for vSphere. This is a must-see session and was a Top10 session (again) this year.

Redefine vSAN Deployments with Next Generation Intel Xeon processors, Intel Optane and Intel 3D NAND SSDs

  • Speakers:
    • Vivek Sarathy (Intel)
    • Kathryn Vandiver
  • Twitter: #STO2705BU

I can be very short about this session: Hardware does matter! Vivek showed us different tests with Intel NVMe drives and future 3D XPoint drives in combination with net Intel XEON Optane processors. If I learned one thing during this session: When designing a vSphere cluster, take a good look at the hardware. The days where you ‘just’ order an server are over!

vSphere Virtual Volumes: Technical Deep Dive

  • Speakers:
    • Pete Flecha
    • Patrick Dirks
  • Twitter: #STO7645BU

For most people, managing a storage array is hard. 
Therefor, I’m a very big fan of VMware VVols. The advantages are enormous:

  • Easier to manage storage (no more LUNs!)
  • Storage policies
  • Faster snapshots
  • PowerShell command for VASA (it’s even possible to script you DR!)

As of vSphere 6.5 array based replication is supported, what was a big gap in the past.
Although the many advantages of VVols, the hardware vendors are not up-to-speed on this. Yes, most of them support VVols. But the implementation how can be questioned.
I sure hope this will improve in the future.

Overall

Every year I’m impressed about VMworld, VMware manage to make it better every year. Even Wi-Fi was working this year (except in the solution exchange).The session rooms are very well setup with an excellent sound setup.

Now, the week after VMworld US, and the week before VMworld Europe, I look back at an excellent week. I had some excellent talks with peers and vendors, great evening with friends (thank you guys!) and learned a lot.

I also will attend VMworld Barcelona. If you like to meet me, give me a ping!

 

 

About Michael
Michael Wilmsen is a experienced VMware Architect with more than 20 years in the IT industry. Main focus is VMware vSphere, Horizon View and Hyper Converged with a deep interest into performance and architecture. Michael is VCDX 210 certified, has been rewarded with the vExpert title from 2011, Nutanix Tech Champion and a Nutanix Platform Professional.

RSS feed for comments on this post.

Leave a Reply

You must be logged in to post a comment.