Helder Almeida - Fotolia
Verizon Cloud will go down for as long as two days for maintenance this weekend, but the real damage may have already been done.
Verizon Communications, Inc. sent emails to impacted customers informing them Verizon Cloud -- the latest iteration of its cloud services -- will be down for up to 48 hours starting Saturday. It's a decision that has drawn sharp criticism as being virtually unheard of in terms of cloud downtime and could scare off potential customers of the nascent service.
BAO Systems LLC, an Arlington, Va.-based provider of security and infrastructure services for federal systems and some non-governmental groups, has been part of Verizon Cloud since it was in beta more than a year ago. Its workloads aren't in production yet as they continue to cautiously evaluate the system performance characteristics, said Kenneth White, a security architect and engineer with BAO.
"When you call yourself enterprise-grade and say there will be downtime of up to two days and have customers shut down VMs and no option for timing, no option for alternate zones or regions, it's frustrating," White said.
White runs systems on Amazon Web Services (AWS), Google Cloud Platform, Rackspace, Digital Ocean, Inc. and Microsoft Azure, but he's never seen a scheduled maintenance like this one. To make matters worse, not only are the VMs going to be down, but the control panel and APIs will also be down, so it will be impossible to have real-time updates on the downtime status, White said.
There are a lot of governmental agencies and groups around Washington that are interested in Verizon Cloud because of success with Terremark and the other legacy Verizon systems, but this shutdown limits options for his clients, White said.
"I would have loved to say when they went general release in August that maybe in February start doing global implementations," White said. "This just pushes it out another six to nine months."
Which Verizon Cloud customers are impacted?
The overall impact on existing Verizon customers will be relatively small, as only 10% of its cloud customers use Verizon Cloud, according to Kevin King, a company spokesman. The remaining 90% run on legacy cloud platforms such as Terremark that host some of the telco's biggest clients.
Kenneth Whitesenior architect and engineer with BAO Systems LLC, based in Arlington, Va.
The maintenance is not related to a third-party bug fix and is "pretty standard stuff" that should be resolved within a day, King said. All affected customers have been notified in order to plan accordingly, he added.
"Updates of this nature typically require some system downtime," King said. "We informed clients to be prepared for the system to be down for 48 hours although we do not anticipate the work will take that long."
It's also not the first time Verizon cloud has gone down for an extended period. In November the cloud was down for maintenance for 24 hours and for 12 hours several months before that.
Cloud outages overblown?
Unlike the Xen reboot that impacted AWS, IBM, Rackspace and others last year, it doesn't appear this will be a rolling reboot with individual VMs down for a period of time, which is reasonable compared to "48 hours where the cloud is obliterated," said Lydia Leong, an analyst with Gartner, Inc., based in Stamford, Conn.
"I’m not sure there's any point in pretending it's routine," Leong said.
IBM shut its pre-SoftLayer cloud down for eight hours citing that customers were largely in test and development, but Leong doesn't remember any major cloud vendor going dark for this long. And that duration is resulting in calls from clients considering Verizon Cloud that want to know what's going on.
"Certainly it's a vast concern to prospects," Leong said.
Despite the concerns, there isn't anything going on here too out of the ordinary, said Carl Brooks, an analyst with 451 Research LLC, based in New York.
"It's completely and utterly normal," Brooks said. "We're back to conflating consumer expectations of always on with cloud infrastructure services, which are a fairly low set of service expectations."
Other vendors do the same thing, they just do a better job of hiding it by doing it in bits and pieces, Brooks said. If anything, Verizon is guilty of setting expectations too low and misreading what non-customers are going to make of it, he added.
It's also highly unlikely that Verizon will hit the kill switch on its data centers, and it would be shocking if the telco was doing some massive forklift upgrade or replacing entire cabinets, Brooks added.
One thing that most observers agree on is that the scope of the impact will be relatively small. When other providers go down, large swaths of the Internet go down, but that won't be the case this weekend, Brooks said.
For some, however, the rocky start for Verizon Cloud doesn't instill confidence that it's ready to handle enterprise needs.
"Today I find it difficult to make an argument to use Verizon Cloud unless you're in a relatively narrow niche," Leong said.
Those reasons could include existing discount credits with Verizon or for companies that want end-to-end Verizon services or have a long track record with Terremark.
One of White's biggest frustrations is that the shutdown meets Service Level Agreements because it's scheduled maintenance.
"I understand large-scale infrastructure systems are one of the most incredibly difficult things that you can take on and I get that this is not trivial," White said. "But if you want to be taken seriously in a market with one, two or three major players, this is not how you do it."
Trevor Jones is the news writer for SearchCloudComputing. You can reach him at firstname.lastname@example.org.