Microsoft Azure customers reporting hitting virtual machine limits in U.S. East regions

Starting a couple of weeks ago, some Azure customers began reporting they were hitting virtual machine (VM) limits in Microsoft's cloud in the U.S. East regions. Users say Microsoft doesn't have enough of the server types needed to spin up certain types of VMs, especially in East US2.

Every cloud vendor sets limits on the various services they offer customers. Microsoft has encouraged users to employ Virtual Machine Reserved Instances, especially in cases where they know ahead of time they will need to scale up. However, the current shortage of Azure VMs seems to be creating worries among customers bumping up against them.

I first got wind of the issues with Azure VM limits from Matt Burcke, who tweeted to me about the his own recent experience. He also pointed to a Reddit thread about capacity issues in East US2. On October 31, he tweeted his response from his support case on VM constraints, which stated:

"Unfortunately, due to high demand for virtual machines in this region, we are not able to approve your quota request at this time. We are continuing to expedite additional capacity for the US East 2 (EUS2)."

The responder also asked whether Burcke is able to redirect his deployments to alternative regions US West 2 and/or start with a smaller number of cores.

The thread on Reddit, which started a week ago, mentions the East US2 capacity limits that some are hitting. One user (BloinkXP) said his/her company was told by their account manager it would take four or so months to get escalation on the issue. More recently he said that the company fortunately got whitelisted, "but it is a datacenter issue and they have over provisioned on ever(y) level. Their stance is 'we aren't sure how we got here.'"

Another user on Reddit, iwifia, said Azure's East 1 was having issues, as well. "They told me to build in West US for things that we need in East. This is making me rethink Azure as a whole and we have been on it for 5 years." 

It's not clear if the capacity limits are hitting other Azure regions, as well. One person in the Reddit thread implied South Central US also might be having issues.

Some users also have been taking to Twitter to report their Azure capacity issues.

In one conversation on the topic, Kevin Beaumont (@GossiTheDog on Twitter) said:

"At Crabbers when they went with Azure they found they couldn't get needed capacity increases on their VMs as the DC was out of hardware for that type, so they had to wait months. ?? it wasn't quite what they were told before hitching their wagon."

He added:

"The Azure stuff is problematic if you do SAP as you need really large reserved instances, what Crabbers found is that what they were sold wasn't actually available after they got dedicated Azure ExpressRoute DC connectivity installed (i.e. financially locked in)."

Waldo Jaquith (@waldojaquith on Twitter) wondered about potential implications around Microsoft's recent Department of Defense JEDI cloud win:

"It's a lucky thing for us all that DOD is nowhere near ready to actually use cloud for nearly anything, because otherwise this would be a disaster. At least this way Azure will have time to ramp up capacity and Northrop et al can learn to use Azure." 

(Kidding -- or not -- aside, Microsoft has separate government clouds and a US DOD East region for the Pentagon's needs.)

I've asked Microsoft for comment on these capacity-limit reports. A spokesperson sent the following reply:

"Due to continuing demand for resources in US East 2, we've placed restrictions on additional quota for some customers and are working directly with customers to provide them with the right options."

I haven't heard any reports of deployment failures resulting from capacity issues, and I'd imagine Microsoft would be giving first priority to existing Azure customers and existing workloads in allocating resources. But it's still something to keep an eye on in the coming weeks and months....