Late Friday evening, millions of people in North America saw a demo of the worst that can happen to cloud computing users. The streaming film service Netflix went dark, halting in mid-movie. At the same time the social networking photo site Instagram went down. These staples of communication and entertainment stayed down, too. Both were victims of a lightning strike on their host facility, Amazon EC2 in Virginia.
The outage was repaired over a span of several hours, and for the most part there was no loss of commerce. Netflix hasn't contacted customers to offer any compensation; Instagram would have no reason to do so, since it's free. But imagine if your cloud-based manufacturing service took a lightning strike. The disaster recovery scenario is significantly complicated when such a key element is outside IT's control.
Amazon's bandwidth for hire has been discussed as a resource for the forthcoming HPA/3000 emulator product that requires no local host. One lightning bolt won't spoil the track record for outside computing services. The new HP Cloud is also bound to weather an outage like this, sometime. However, taking hosting virtual as well as remote/offshore means reworking disaster recovery concepts. When relying on the cloud to run manufacturing, a rapid cutover capability to another provider could save millions of dollars in lost operations.
It could also save a manager's job. On Infoworld's website one of the most popular stories from June was "Adopt the cloud, kill your IT career." The point is not that cloud computing is less stable. Rather, "It's irresponsible to think that just because you push a problem outside your office, it ceases to be your problem." Since the start of 2012 Kenandy Inc. has been offering a replacement for HP 3000 MANMAN software, all based in the cloud. Its high-level answer about a cloud outage problem has been an interesting part of this kind of transition: We know redundancy. Regardless, salesforce.com experienced an outage Thursday, less than 48 hours before the Amazon lightning strike. A little under five hours of downtime ensued.
Salesforce is an equity partner in Kenandy, and there's no mention of using Amazon's cloud services in company presentations. Salesforce.com has had other outages in the past. Reports show that the operation is centered in a single Silicon Valley datacenter with data shadowed to another facility on the US East Coast. More than 70,000 customers count on the stability of salesforce.com.
Kenandy calls its product the first cloud ERP built entirely on salesforce.com’s social enterprise cloud computing platform, specifically for product companies. In May, Social ERP added financials and order management to the manufacturing management core. The prospective customer is more than just MANMAN sites. The target is companies that design, manufacture, and distribute products, so they can control and get visibility of their supply and distribution networks.
“With the addition of financials and order management, Kenandy Social ERP becomes the backbone of the social enterprise,” says Sandra Kurtzig, Chairman and CEO of Kenandy. “It’s time to re-think ERP, and that’s what we’ve done. Kenandy release 2.0 now offers fully integrated end-to-end ERP and it’s entirely on the cloud, easy to use, fast to deploy, mobile, global, and social.”
“Kenandy Social ERP gives our customers the ability to transform into social enterprises across both the front and back office, entirely in the cloud,” said Ron Huddleston, senior vice president, ISV and Alliances, salesforce.com. “With the rich set of add-on apps in the AppExchange and user extensibility through Force.com, companies are only limited by their imagination.”
That Virginia lightning strike could just as easily been a hammer thrown onto a single company's datacenter, or even upon a network service provider that links hosting to the rest of an enterprise. The cost savings in cloud computing go beyond elimination of hardware by moving it into the cloud. For $175 per user per month at Kendandy, you get your share of access to IT staff which won't pay to hire exclusively. But there's little you can do in the event of a problem except call that staff -- just as a half-million East Coast electric customers did starting after Friday night's storms. By Monday morning, 80 percent of them were still without power. As the InfoWorld article states
You're adding another avenue for the blame to follow. The end result of a catastrophic failure or data loss event is exactly the same whether you own the service or contract it out. The difference is you can't do anything about it directly. You jump out of the plane and hope that whoever packed your parachute knew what he or she was doing.
A company can't expect to be able to hire subject experts at every level of IT. In this view, working with a cloud or hosted service vendor makes sense because there's a high concentration of expert skill at a company whose sole focus is delivering that service. There's some truth to that, for sure, but it's not the same as infallibility.
The HP 3000 homesteading customers who are some of the best prospects for using cloud computing are those trying to trim IT budgets. They'll need assurance that the cloud providers of ERP, CRM or financials have those experts on call, as well as a backup set of servers -- not just data -- which are well-separated from bad weather.
Clouds turn out to be just as susceptible to weather disasters as in-house IT. The cautions which the 3000 customers have voiced so far might stem from the out-of-house recovery that the cloud demands. This has always been a belt-and-suspenders community. But that's an old-school expense that can seem less costly after a dark and stormy night, one when the movies flicker to a halt.