We’ve been following Eucalyptus for some time, and they recently invited us to a briefing about a new alliance called NRE, which is a credible group of independent vendors, newScale, rPath and Eucalyptus.
This wasn’t spun from an Open Source prespective and it was interesting to see the Eucalyptus positioning to the general marketplace. Eucalyptus is positioned as the “leading” Open Source cloud, the benefit of Open Source being it is “on your own terms”. It offers IAAS in the data center, just like Amazon Web Services. It is Elastic, based on industry standard APIs, hypervisor agnostic, supports both Windows & Linux guests, and has a huge ecosystem. It’s the elasticity and the scalability that are driving the adoption. Pricing is secondary, and you also get the feeling that it’s not traditional enterprises which are picking it up.
As far as the broader Cloud picture is concerned, like many analyst-led vendor briefings it seemed that the actual premise wasn’t to be discussed. It had that “presumed sale” flavor beloved of car salesmen, “when you buy the car you will experience…”, not “this is why you should buy the car”. The assumption was that we were all on the yellow brick road to the Cloud, and that Cloud itself was across the chasm and in to the “early majority”. To back this up there was Forrester data that 25% of larger organizations would like to build their own cloud in 2011, with smaller organizations interested in public/hybrid options.
So, when you buy the “Cloud” (at least from these vendors) you will experience
- self-service (newScale)
- deep automation (rPath)
- elasticity (Eucalyptus)
- and cloud-bursting – through … standards (note the “s”)
It’s a good solution, the bits make sense together, and they are all credible vendors.
But bear in mind, Cloud is not a product. It is a change in how you operate IT resources, out with the old, in with the new. No room for ITIL zealots here, no “cloud-washing” your VMware datacenter (i.e. claiming it’s a cloud when it sooo obviously isn’t), we are all on a 5 year journey to private cloud.
However as we progressed through this briefing we did at least start to understand the rationale and value proposition, such as it is.
1) Fear that has come from transparency in pricing of public cloud. If Amazon can do it for this price, why does IT do it for,… errm, actually we don’t know how much IT costs but we think it is quite a lot.
It’s strange, because it turns out there are things that IT does in terms of security and governance that are slightly different, so there’s no apples-for-apples comparison, and there’s no real chargeback (we’ve been on an “journey” to that for at least my 20 years in the business), so IT isn’t metered, and private cloud becomes a rod to beat up IT over price, in some ill-defined way and perversely, a shield that IT uses to deflect such attacks – Cloud saves money (because Amazon appears to be cheaper), and in an unmetered IT shop, who’s counting anyway?
The bottom line is that if we sell Cloud on point 1) way we’re missing the first rule of “enterprise sales” – never sell on price.
2) Application agility. Somehow the cloud allows you deploy applications faster through a self-service portal, but when you drill into this the usecases (or beach-heads) are the usual suspects, development, test/QA. When it comes to the Business deploying applications that way, someone forgot that the application actually has to be written, or bought and configured, which does tend to take longer than rolling out a VM to run it on. And if the CMO wants something quick and dirty in the Cloud, by all means let him get someone to prototype it there and IT can bring back that “public cloud leakage” into the vmware datacenter (whatever that may cost) if and when it becomes mission-critical.
3) Elasticity. There is clearly some benefit as Business Units can be persuaded to pool resources, but the dominant costs software, with predominantly user-based pricing, and management, where there are as yet unproven benefits over straight virtualization, power and real-estate and hardware do contribute, but not really enough to justify the Cloud investment. In any case unless you can cloud burst, this is elasticity within limits, the limits of physical hardware, or quotas.
So, without cloud-bursting, there isn’t anything so far that really stands up to scrutiny, but we did at last come to something that made sense and where the Cloud demonstrably does it better. That area was governance. The idea is that through the level of automation built into the cloud it becomes possible to ensure architectures are automatically constructed which meet certain policy guidelines – access is provided only in certain ways and between certain machines, or data is physically located in certain geographies. It’s easier to do this in a dynamically-configurable environment than a manually-built one.