Cloud platform API spec on track, due September
Six months on, we catch up with the authors of CAMP, a cloud management API being drafted by Oracle, Red Hat and Rackspace
The CAMP (Cloud Application Management for Platforms) specification defines an API for interacting with cloud platforms, and is being thrashed out by a cross-industry group under the guidance of not-for-profit group OASIS.
Last time JAXenter spoke to the people behind CAMP, the spec had only just been publicly revealed with minor fanfare. Yet six months down the line, press coverage has entirely disappeared.
The good news? So far, the spec has been “progressing very, very smoothly”, says Martin Chapman, Oracle standards expert and chair of CAMP’s Technical Committee. “We have resolved 30 technical issues that have been raised since the beginning of the committee.”
This committee includes representatives from Rackspace (represented in this call by Senior Technology Strategist Adrian Otto), Cloudsoft (CEO Duncan Johnston-Watt and CTO Alex Henevald are on the line), Red Hat, Software AG and WSO2. More recent additions to the TC’s 40 members include Google, Fujitsu, Huawei and the US Department of Defence.
Of technical issues sorted so far, most have been “low-level [operations], how to do REST and JSON, a lot of religious debates – no, sorry, lots of potential religious debates,” says Chapman. “In fact the TC has been very amenable to debate and resolving all the religious aspects of REST and JSON.”
The latest draft [1.4 MB PDF], published earlier this month, is so different from last August’s initial release that “you might not recognise the two, sitting side-by-side”, says Otto. New features of the API include use of HTTP PATCH, discoverability of different serialisation formats and – best of all, they say – extensibility.
The TC’s biggest quandary was to “make an API extendable so that people can do things in a structured and standardised way, but still have the freedom to do their own flavour of something”, says Otto. The TC were so proud of their eventual solution (outlined on page 46 of the current document) that “there was actually applause when we carried the motion”.
Specifications tend to take a notoriously long amount of time to be completed; the HTML5 spec (or at least one version of it) took six years to be fully written. In contrast, CAMP is hoped to be finished by Autumn, with less than three years’ work under its belt.
“As it currently stands we are halfway through our issues list, which is actually quite remarkable in such a short space of time,” says Chapman.
“In the standards world, this is moving at Mach 2,” agrees Johnson-Watt, emphasising the need to get the spec down “before the genie’s out of the bottle”.
But that genie may already have escaped. A recent report by Synergy Research Group claims that the top three PaaS providers worldwide are current Microsoft, Amazon and Salesforce – none of whom have yet to show any public support for the CAMP specification.
When JAXenter approached the market leaders for comment, only Salesforce replied, confirming that while the company is not currently implementing CAMP, “we are always evaluating standards/specs and never rule out involvement”.
“Amazon doesn’t seem to be playing much in the cloud standards space,” says Chapman. “I mean, they do play in the lower-level standards, but in the cloud standards they seem to be very absent at the moment, and we keep trying. We will keep on trying.”
What the spec lacks now is feedback from end-users, and a 30-day public review is “the next major step”. “I think we are scheduling that for early summer,” says Chapman.
After that, a first version of the spec will hopefully be complete by September, and the API will be ready for implementation. However, even those on the board are somewhat coy about exactly how CAMP will be supported in their companies’ products.
Chapman says that Oracle are “prototyping and implementing the spec as we go along” while Henevald says that Cloudsoft “will have something, I think, this summer”. Otto, meanwhile, admits that while Rackspace has “platforms and services that could potentially use the spec,” making it useful is “kind of a catch-22 – it’s much easier to have the spec done and then have people begin implementing it at that point”.
Still, with so many parties already involved in the spec there’s reason to be positive. If CAMP is adopted even by just the industry’s minor players and proves a success, it may put pressure on the big boys to jump aboard. It’s worth floating the idea, anyway.