VMworld 2017 Roundup: Day 2

VMworld's first full day of sessions and such delivered. Here we go.

General Session

For general session commentary, check out the Live Blog. To sum it up quick:

• VMware Cloud on AWS was officially released,

• VMware's betting big on NSX by pushing it beyond on site and cloud data centres, extending it to containers, app platforms, and more, and

• VMware introduced AppDefense, which is their new security play.

Meet the vCenter Server Experts Panel [SER1440PU]

The first session of the week for me was a Q&A panel, featuring Mike Foley, Adam Eckerle, Kyle Ruddy, Dilpreet Bindra, and Emad Younis. They started things off by mentioning that the vSphere web client is going away, with a sun setting announcement that was delivered last week. vCenter Server on Windows is also going to follow the same path, with the VCSA standing as the successor. There hasn't been an official date set yet, however it's expected that these products will be retired as of the next major release of vSphere. If you're still on vCenter Server for Windows, be sure to look at the migration tool as it will help you seamlessly replace it with VCSA.

There was a big of panel discussion among themselves, highlighting some vSphere 6.5 benefits (such as VM encryption, secure boot), the ongoing transition from SOAP-based to RESTful APIs, and that PowerCLI modules are being evaluated and re-written to make them better and faster (case in point, Get-Vm). It's recommended to run the latest PowerCLI version and not stick with the version of PowerCLI that matches your vSphere deployment, as these kinds of improvements can often be "retroactively" available to older vSphere environments, since it's the PowerCLI code itself being improved. Also, check out the VMware API Explorer to browse, search and inspect the various VMware APIs.

The audience was then invited to pose questions to the panel:

• Q: Will UUIDs be accessible via APIs?

• A: That's coming in the future; VMware is interested to know what use cases customers have for the UUID attribute so they can better understand the need and deliver.

• Q: Will vCenter be able to be upgraded via VUM?

• A: That functionality's being worked on.

• Q: VAMI can manually backup but not be scheduled at the moment, is that coming?

• A: Scheduling can be handled externally (ex. via PowerShell) at the moment, but it is being worked on.

• Q: Is VMware getting rid of the PSC? Certificates are especially difficult to manage.

• A: The number one use case for external PSC deployment is the need for vCenter Enhanced Linked Mode. VMware is working on making Enhanced Linked Mode available with embedded PSCs. In the meantime, if customer's don't need Enhanced Linked Mode they should stick with embedded PSC.

• Q: Is high availability (HA) being brought to PSC?

• A: For external PSC the official guidance is to use an external load balancer. If you use the embedded PSC it can be covered by vCenter HA.

• Q: Can vCenter warn is the PSC isn't at the right version before vCenter itself is updated?

• A: This exists in 6.5 already.

• Q: Can SSO domains be merged (for those customers who have made mistakes, for instance)?

• A: It's being worked on for 6.x. If you're still on 5.x then vCenter can consolidate SSO domains before the upgrade to 6.x. Once a vCenter component is upgraded to 6.x, however, the SSO domains are "locked in".

• Q: Why is FT limited to 4 CPUs? Can it be increased?

• A: The limit is based on engineering considerations, however increasing the FT maximums is being worked on. No dates can be provided at this time.

• Q: When will the new vSphere Client be feature complete?

• A: It's now about 95% feature complete with the 6.5u1 release, and VMware is working on reaching 100% feature parity with the vSphere Web Client.

• Q: Are there any plans to increase the maximum cluster nodes for DRS (currently at 64)?

• A: It's in the works.

• Q: Some Auto Deploy features are still PowerCLI only, will they be added to the vSphere Client?

• A: It's on the road map.

• Q: Is certificate installation & management being improved? Not all companies can make vCenter an intermediate CA.

• A: That limitation is understood. A hybrid approach is recommended where the certs are replaced on the PSC and VCSA, and then those components are allowed to mange the certificates installed on the hosts. This secures access to vCenter itself, which would be needed to attempt to get to the host interfaces anyway.

The panel ended with a quick note that vCenter HA is not the same as vCenter DR, so plan for both.

VMware Cloud on AWS - Getting Started Hands-On Lab

Due to a scheduling change, I found myself with a few minutes to kill, so I decided to visit the Hands-On Labs. I sat the VMware Cloud on AWS - Getting Started lab which conceptually explained the service and then walked you through the provisioning of an SDDC with associated network configuration (what could be accomplished within the confines of the HOL environment, anyway). I was struck by how straight forward the provisioning process was. If you plan out your VMware Cloud on AWS environment carefully before hand, accounting for the network requirements such as VPN, firewall rules, IP ranges, etc., then you can have your environment up and running really quickly.

CONVERGED Meeting

CONVERGED is the user group for the DellEMC converged and hyper-converged product lines. They had a meeting at VMworld for existing user group members as well as people who may be interested. The meeting started off with Michael Dell fielding audience questions, such as how Dell Technologies is doing now that it's been a year since the acquisition of EMC. He mentioned that the biggest surprise is that they really haven't encountered any big surprises, but instead have found that there have been unexpected synergies between the companies under the new Dell Technologies umbrella. Another audience member asked about Dell Tech buying back VMware stock, and Michael acknowledged that they are indeed buying some of both the regular VMware stock as well as the tracking stock. My guess is that this is a general tactic meant to improve their investment and control standing with VMware. It remains to be seen if Dell Tech may try to buy back the remaining stock to take VMware private. It seems to be working well enough in the state that it's in now.

When asked about Vblock and VxBlock, Michael Dell said that they are unique products and that Dell isn't pushing to have Dell hardware used in those offerings. This is due to their existing partnership with Cisco. If customer's want Dell EMC hardware-backed (hyper)converged solutions, Dell EMC has other offerings. When asked about cloud and future offerings, Dell shared that his view is that cloud is not a place, but a way of doing things. The current focus will remain making on-premises systems work and perform better for customers with a goal of delivering competitive pricing. He noted that one of Dell EMC's customers realized something like a 4x-8x savings remaining in on-prem private cloud versus moving their workloads to a vendor's cloud offering.

After Mr. Dell left, a systems administrator from Johnsonville Sausage then had an on-stage interview about their use of converged systems. Johnsonville had implemented Vblocks for their ERP and other services (SAP, SAP HANA, time keeping, etc.). Last year they implemented VxRail for specific use cases, for instance deploying infrastructure to their international locations in 30+ countries where they wanted more of a hands-off approach to the infrastructure. It was interesting to note that Johsonville leases their hardware and that only two administrators look after their entire infrastructure, which averages 10-17 VMs per appliance. Converged infrastructure has allowed them to do platform upgrades during business hours in about 45 minutes with no interruption to business services.

Chad then took the stage with Ray O'Farrell to take audience questions. Those who have experienced Chad in person before knows that he is an energetic and enthusiastic speaker, so I'll summarize the points I was able to make note of:

• "The buck stops with me (Chad) for all things converged."

• The net promoter score for Dell EMC converged infrastructure (CI) is high, meaning customers are happy in general. Chad acknowledges that that doesn't mean that there aren't customers experiencing issues and dissatisfaction.

• Reducing operational complexity by making infrastructure upgrades easier is a hard promise to keep due to the nature of the complexity itself, however Dell EMC is striving to do just that.

• "HCI is not the way forward for everything"; different infrastructure may be needed for different workloads, not everything's the same.

• vSphere versions for (H)CI solutions lag behind the vSphere general release because it has to go through extra validation and testing. (H)CI customer's typically value the stability of the solution over being on the bleeding edge. Those that want/need the latest code should not go with (H)CI.

• They're trying to use UI releases as the primary release gates with (H)CI, so it may seem like the updates don't come up with great frequency.

• "Upgrading vCenter is one of the harder parts of a vSphere Upgrade"; "vCenter is often the root of (H)CI upgrade failures".

• Chat was going to unilaterally axe the Cisco Nexus 1000v virtual switch in the (H)CI products due to the catastrophic failures some customers were having and the low adoption rate of the product.

• A customer voiced a concern about migration paths away from the Nexus 1000v as Dell EMC did not offer any prescriptive way forward. Chad responded by telling the customer that he had been heard and will take the concern back to the team to see if there's a way to help smooth the transition away from the 1000v.

• VxRail with vSAN has been a hit for Dell EMC.

• The combined companies will be taking a different direction than was shown at yesterday's (Monday) VMworld general session. He was likely referring to the announcement of Pivotal Container Service (PKS); see our Tuesday VMworld General Session coverage.

Hackathon Training: Getting started with the vSphere Automation SDK for Python [VMTN6720U]

I applaud the intention here, to deliver a quick introduction and overview related to coding against VMware components, however my experience has been that these efforts are often plagued with unforeseen issues if you try to rely on attendee's devices. Such was the case here, where I could not get the Python examples working on my laptop until near the end of the session since I ultimately discovered that my Python environment variables were not set properly.

I'd recommend providing some sort of predefined environments for attendees to try the examples in the future. With Python, for instance, Jupyter Notebooks could have been provisioned ahead of time so that the instructor had a predictable, consistent set of environments to offer. Otherwise the training went well, and I was actually able to carry out the core activity, which was retrieving information from and manipulating a vCenter environment via Python.

One small take away was the introduction of Postman, which is a graphical API development environment. This will let you easily explore an API to understand it better and test it out before incorporating it into your development.

Hackathon Event: 15 teams hack on ideas! [VMTN6722U]

The last event for the evening was the anticipated Hackathon. I participated on team 8: Need-for-Speed. Our goal was to find and make performance improvements to the vCheck-vSphere script. Unfortunately the event was plagued with wi-fi issues, so we were unable to spin up a validation environment and commit changes, however our team was not the only one affected. We did, however, manage to get some code together that should introducing parallelism to the vCheck plugin execution. In theory this should deliver a performance boost as the plugins would run in parallel, four at a time, and not run one-at-a-time serially. We hadn't yet worked out how to find and address plugin dependencies (especially the first few plugins which set up the connection to vCenter and define some commonly referenced functions), but that can be worked on in the future. We were also able to find an inefficiency in a common function that retrieves events from vCenter. The function was not accepting an event category parameter, which some plugins were expecting. Fixing this gap and updating the relevant plugins should improve query filtering, which is much more efficient that filtering post-query.

All in all, our team had a great time together working on the project, even if our accomplishments weren't terribly tangible. This event is still on track to remain my favourite at VMworld, and I'm looking forward to next year!

In the Can

That's VMworld 2017 US Day 2 in the can. The Day 3 general session live blog coverage is up and a Day 3 recap on the way, stay tuned!