#barb darrow

Verizon's cloud was offline for more than 40 hours over the weekend. 

The company warned customers that this maintenance period would take place, and argued that it was necessary to set the stage for bigger, better and more stable things to come.

There was quite a lot of (almost justifiable) grumbling that clouds shouldn't go offline for almost two days, and that it should be eminently feasible to upgrade bits of a cloud while (seamlessly and invisibly) moving customer workloads to other machines. The underlying hardware could, it is often argued, be completely replaced, while the cloud - and its customers - continue to work. That's certainly the hype, but Verizon's (and other) outages demonstrate that it's not always the case.

Maybe Verizon could have carried out this maintenance over a longer period, at greater cost to them and with greater complexity, but with far less impact on customers (or market perception). Or maybe they really did have to shut the whole thing down for two days. I suspect there are equally vehement, dogmatic and 'fact'-filled ways to argue both cases.

Now the cloud is back, Ben Kepes is one of those to argue the extended maintenance window has done some good:

"Verizon’s outage, while unfortunate, is set to introduce key new functionality. As such it would seem to be a positive development for customers"

Over at Gigaom, Barb Darrow seems less convinced:

"Last weekend, when Verizon advised customers a week in advance of  what it said could be a 48-hour shutdown for planned maintenance, all sorts of things hit the fan. The prevailing opinion was that cloud computing vendors should be able to handle upgrades and maintenance with a lot less downtime than that."

Whether necessary or not, I can't help feeling that Verizon could have done a far better job of communicating with customers and the world during the outage. 

The number of "still down, still no news" tweets during the outage was bad for Verizon, bad for their customers, and bad for cloud.

Trying to work out who's winning (and by how much) in the public cloud space is, reckons Gigaom's Barb Darrow, a 'Mission Impossible.'

Over at Gigaom, Barb Darrow reports Salesforce CEO Marc Benioff's announcement that

[Salesforce uses Oracle] but, [Benioff] added, that doesn’t mean it’s not using a ton of other stuff

The tech media, it seems, loves these stories. PayPal uses OpenStack. NASA doesn't use OpenStack. Microsoft uses Macs. CERN uses whatever CERN's using this week.

And on it goes. And the implication always seems to be that the arrangement is an exclusive one. If PayPal uses OpenStack then (surely?) it must not be using any of the various alternatives.