IBM killing off its first go at cloud object storage for Spark – 30 months after launch
Move your data by August 24th or lose it, then ask if this would happen on-prem
We all know cloud is evolving fast, but IBM's just given us the downside of that speed: a service it switched on in February 2015 will be switched off in August 2017.
That service is the first iteration of Bluemix's Swift-based Object Storage for Spark as a Service users.
IBM's now on version 3 of the service and that's where it wants users to migrate. “We will now be deleting all existing instances after 30 days i.e. on August 24, 2017,” IBM says. “We recommend users to unprovision the Object Storage v1 service and switching to v3, before August 24, 2017,” the company advises.
Object Storage V1 for Spark didn't last long – it was restricted to “private” status in February 2016 after IBM tried to style it as more of a beta despite the v moniker. V3 of the service arrived not long afterwards and remains “available and accessible for use”, IBM says.
IBM's guidance for users suggests all you need to do is copy data from V1 stores to V3 stores, which doesn't sound onerous.
But the job still has to be done and is needed just 20 months after launch. IBM users may be getting used to this kind of rapid forced upgrade: the company recently set a 60-day deadline to adopt an updated active deploy tool.
That's a situation we cannot recall ever having faced users of on-premises equipment. Which of course isn't easily elastic like cloud storage, nor can it easily tap the on-demand Apache Spark that IBM offered alongside V1. But IBM's fast upgrade cycle is also a little different to the typical SaaS experience, in which platform upgrades are typically non-disruptive. And of course betas of anything are generally close-to-final versions of a product, not a soon-to-be-superseded preview.
IBM argues that V3 is superior to its predecessors, thanks to the addition of "an enhanced authentication model, UI and self-provisioning improvements for a better user experience" and is therefore worth the move. It needs to be an improvement: Gartner recently opined that Big Blue's cloud “has not improved significantly since the [SoftLayer] acquisition in mid-2013; it is SMB-centric, hosting-oriented and missing many cloud IaaS capabilities required by midmarket and enterprise customers.” ®