On Black Friday, Mason Garments’ online store experienced severe performance issues. Their previous server at a different hosting provider, normally perfectly adequate, simply couldn’t handle the enormous numbers of visitors that day. We decided to pull out all the stops to get them back in business again. With success. Within seven hours, we had them up and running on our largest Hypernode setup, and the orders were streaming back in.
Panic City: Down on Black Friday
Friday morning, 8:30 a.m. It’s Black Friday. “Hi, this is Mike from Schluter IT. I’m the developer for Mason Garments. It’s Black Friday and our online store has gone down, our hardware isn’t up to it. Is there anything you can do for us?”
Their Magento store was located on a dedicated server somewhere else, which had failed to survive their first Black Friday mailing on Thursday afternoon. The store was still reachable, but agonizingly slow, and worse, no orders were making it through at all. We repeat: it was Black Friday…
Mike had spent hours attempting to get the store up and operational again, in vain. He’d set up, configured and rolled out a new server, but the capacity was still insufficient. The other hosting providers he’d contacted were no help. It would take them days, and who has time for that on Black Friday? That website needed to be back online, by whatever means.
“You’ll be online by this afternoon”
“Sure, we’ll make the time,” Account Manager Ilana said. “Let’s go for it. We’ll have you up and running by this afternoon.”
She could promise that because we can do emergency migrations much, much faster than other hosting providers. They must first free up and configure a server, whereas we can have a fully configured Hypernode up and ready to use in under twenty minutes.
Ilana set up a Slack channel with Mike and Rens, the Hypernode Support Engineer. They defined a clear roadmap (including items such as how to save more time by skipping any non-essential parts of the onboarding process), Mike gave Rens access to the server and the migration began.
Mike Schluter: “The communication process was remarkably smooth, with plenty of clarity and strict agreements. That inspired confidence.”
In twenty minutes, a Hypernode awaited Mason Garments’ store. This store proved to contain various custom configurations, which Rens converted to the Hypernode configuration. He also fine-tuned Varnish to enable efficient processing of the enormous loads.
It was soon obvious how woefully undersized the old dedicated server had been for Black Friday. The old server had 8 CPUs, and we spun up a somewhat larger Hypernode for the migration. We got the store back up, but the performance still left something be desired. Without delay, we launched our next migration, this time to a Hypernode on AWS, with 128 CPUs and 1952 GB of memory.
Black Friday Campaign Finally Launched
At 3:30 p.m., seven hours after Mike’s decision to call Hypernode, Mason Garments’ store was running like clockwork again and the orders were flooding back in. The Black Friday campaign they’d prepared for so carefully had finally begun. Within three hours, the store had processed six times as many orders already as during the entire 24 preceding hours. The store was handling 1000+ visitors at a time without a hitch.
* Source: MageReport Premium – magereport.com
Mike Schluter: “The fact that Hypernode lets you scale up or down easily is a huge bonus. You can respond appropriately, even when e-commerce visitor numbers are capricious. They know what they’re doing!”
So What Went Wrong?
Mason Garments’ dedicated server had insufficient capacity to handle their enormous numbers of visitors. It happens. Even on a Hypernode.
And when it does, the important thing is to have a procedure in place to get the store up and running again right away. In this case, there was no way to scale up the capacity, because the dedicated server was also a physical server.
At Hypernode, that’s very different:
- A Hypernode isn’t a physical, dedicated server, but a Virtual Private Server (VPS), ‘in the cloud’. A simple click is enough to create a new server.
- We use auto recovery; if our monitoring notices a Hypernode (VPS) is experiencing unresolvable issues, it automatically starts up a process to copy the store to a new Hypernode.
- Because we employ a standard configuration, the new Hypernode is configured and ready to use in under twenty minutes. No human intervention is required, nor days of work.
So, should Mason Garments ever be caught off guard by their own success again in future: we’ve got your back 🙂