filograph - Fotolia
In a bit of irony this week, Verizon said its nearly two-day outage that generated considerable public backlash was intended to keep this very thing from happening again.
The weekend outage was tied to an upgrade to Verizon Cloud for "seamless upgrade functionality" that would allow for major system upgrades without service interruptions or limits on infrastructure capacity, Verizon said in a press release. The statement came after a 40-hour downtime -- originally predicted as lasting up to 48 hours -- that ended on Sunday at 5 p.m. EST.
One of the more vocal critics of the outage on Twitter has been Kenn White, a security architect and engineer with BOA Systems LLC, a company based in Arlington, Va., that provides security and infrastructure services for federal systems and non-governmental groups. White has been testing workloads on the platform and said in an interview that Verizon's claims of being enterprise-grade don't jive with this level of service disruption.
As a result of the upgrade, customers won't need to set up virtual machines in multiple zones or upgrade domains, nor will they have to reboot VMs after maintenance is completed, according to Verizon.
But that doesn't give much solace to White.
"It's an interesting claim, but people don't make these kinds of business decisions based on marketing claims," White said. "The only thing that matters in this space is performance."
It would be helpful to learn about some of the technical aspects of this latest improvement, White said, adding that he's ruled out any production use of Verizon Cloud for at least the next year.
"The thing that's sort of odd about this is Verizon has an incredibly talented engineering team," White said. "They could be sitting on some really innovative technology, but I have no idea because they've communicated as ham-fistedly as possible."
Verizon's outage is déjà vu all over again, said Carl Brooks, an analyst with 451 Research LLC, based in New York. Amazon got the same kind of complaints four to five years ago, as did Microsoft when Azure went down in November, but new customers continued to sign up, he added.
"From a PR perspective, it's yet another object lesson in how to manage expectations around IaaS, which we've seen many times before, and that each provider apparently has to learn for itself," Brooks said.
Live migration and applying patches down to the firmware without taking out an entire rack are things that Amazon, Google and Microsoft have been able to do for a long time, but it's still a hurdle for many of the lower-tier cloud providers, said James Staten, an analyst with Forrester Research Inc., based in Cambridge, Mass.
The maintenance outage didn't impact Verizon's legacy cloud products, including its Terremark platform. A Verizon spokesperson declined to answer questions about the outage this week or make company officials available for comment. It's unclear why Verizon didn't make the upgrade during the beta period, which ended late last summer, but it's possible the technology wasn't available at the time and the vendor was under pressure to get the platform into general availability, Staten said.
Still, the impact of the outage is probably pretty low, he added.
"If you're going to do it in [general availability], do it early," Staten said. “Only about 5% of their other services have moved over and, from what we can see, most of the new cloud customers aren't in production."
Trevor Jones is the news writer for SearchCloudComputing. You can reach him at firstname.lastname@example.org.