Hitless WLAN Controller Upgrade?

The mythical hitless WLAN controller upgrade. Something that under no circumstance is possible, even if we’re only talking about the end user experience….right? A prime example of why controller-less wireless networks are the only way to go….right? In the words of ESPN personality Lee Corso, Not so fast!

With the release of ArubaOS (AOS) 8, a couple of new features were introduced to enhance the wireless network, clustering and Live Upgrade. (For those of you familiar with Aruba, a Hewlett Packard Enterprise company these features were pitched to the world at the Atmosphere 2016 conference at the Cosmopolitan in Las Vegas.) The leaders at Aruba felt that it was time to truly treat the wireless network as the mission critical infrastructure that it has become to end users. With many customers, it has become a fact of life that a network upgrade will be a tedious process, one that requires immense planning and advanced notice to their respective organizations in order to complete.

Recently with the release of AOS 8.2.1, I had the opportunity to really put the live upgrade feature to the test in my lab. My lab setup is a mobility master virtual appliance running on an HP z820 workstation (my lab VMWare ESXi server), with two Aruba controller clusters (one physical cluster: 1-7005 & 1-7010 controller / one virtual cluster: 1 MC-VA-50 & 1 MC-VA-250), and 4 access points (2 AP-325, 1 AP-303H, and 1 AP-214). Off of this infrastructure I have approximately 20 devices connected that are a mix of IoT, laptops, tablets, etc that were all online during the upgrade process. It’s important to note that some of the IoT devices are thermostats that are extremely quick to send out email alerts if they lose connectivity, even for a short period of time. Typically a controller upgrade in the past would defiantly be long enough to issue a few of those email alerts. With the basics out of the way, time to dive into the cluster upgrades!

The initial upgrade was of my virtual cluster as this is really a testbed for things, and nothing of critical importance resides there. Once the code had been downloaded to my local FTP server, the upgrade process was initiated on the mobility master with a few bits of info input (code version, FTP IP address, etc). Much to my surprise, after just a handful of clicks the process was pretty much a “sit back and relax” process for me. Once the process completed (time will vary based on virtual server environment) I was very pleased to see no errors and decided it was time to bite the bullet and complete the physical cluster upgrade.

Screen Shot 2018-03-20 at 11.48.58 AM

Upgrade Status from AOS 8 Mobility Master

The process to complete the second cluster upgrade was expedited slightly, by the ability to use the same cluster upgrade settings used previously. ArubaOS 8 allows the user to save the upgrade parameters into an upgrade profile, that allows quick future upgrades if the variables remain the same. Just prior to kicking off the upgrade, I started a continuous ping to google.com from my MBP attached to one of the AP-325 joined to the physical cluster. The process again completely flawlessly, without any user intervention required. After the completion of the upgrade process, the continuous ping reveled a mere 20 pings dropped the entire time (out of 2843 packets) and most of those on this device were due to poor overlapping coverage in the test area. The most astonishing thing is that during the upgrade, I was able to continue to perform work without ANY interruption on other devices I had online. This was truly impressive.

Screen Shot 2018-03-20 at 12.00.12 PM

Total Packet Loss during upgrade (to google.com)

Let me take it one step further. Without even realizing this until the upgrade was complete, I had an Apple TV streaming a live college baseball game throughout the ENTIRE upgrade process. Not a single piece of live game action was missed, not even a blip of the video or audio.

A true hitless upgrade to the end user device, obviously not. A host of things happen in the background such as roaming to a new AP, buffering on the Apple TV, etc. Yet none of these events took long enough for the end user to notice. The fact that I could continue to work and stream live video without any interruption in service from the wireless network, is spectacular. This may be as close to a true “zero downtime” upgrade as possible. So Aruba, kudos to you for providing your customers with the tools and features within ArubaOS 8 to legitimately create a mission critical wireless network, with unparalleled performance and resiliency.

For more information on how the ArubaOS 8 Live Upgrade works, check out this Airheads Community post for details.

Questions/Comments as always are welcome.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s