Is there a chance of cloud sanity breaking out?
Thu 10 Jul 2014
Most major vendors have agreed it is good to avoid locking in their customers and now the discussion has moved on to realising the promise of the cloud. So, now, Clive Longbottom looks forward to a time when the composite app becomes the norm
At one stage, it looked like the cloud computing community was hell bent on destroying itself before the market had even taken off. Amazon Web Services (AWS) showed the way with its own offering and the rest the tried to follow. Google and Microsoft came out with their own clouds, and then each of the main technology vendors had various attempts at creating their own platforms as well.
It looked like there would be masses of proprietary cloud platforms proliferating, with any company daring to use one finding itself locked-in to that specific platform. Research by Quocirca found that many users put off any major cloud plans for this reason – moving from in-house lock-in to an outsourced lock-in did not make much sense.
However, it now looks like a market where things are calming down. Where infrastructure as a service (IaaS) platforms from the big IT vendors such as IBM SoftLayer or HP Helion, are supporting the overlaying of a standardised cloud platform, such as OpenStack. Where a cloud platform as a service (PaaS) is being offered (such as Rackspace – the first driver behind OpenStack – or DreamHost), it also seems to be gravitating towards OpenStack. Where the provider has already invested too much to change direction (as is the case with the Big 3: AWS, Google and Microsoft, APIs and standards are emerging that should allow for relatively good levels of interoperability between clouds at the workload and process levels.
This is important because of two main reasons.
Firstly, no-one should be locked in when it comes to choosing an external provider. That provider could go bust; they could be acquired by a company that you do not like. There needs to be a capability to move workloads and associated data quickly and efficiently from one cloud to another.
Secondly, the promise of cloud computing is not in moving monolithic applications to another place and running them as is. All that this does is replace internal hosting with external hosting: this is not, in Quocirca’s thinking, true cloud computing.
No, over the longer term, there needs to be a breaking up of monolithic applications to create a composite app. A composite app takes sets of functions and integrates them to build a dynamic application that facilitates a given process at a particular time.
For example, you may want to make an offer to a specific group of customers today. Tomorrow, you may want to make a completely different offer to a different group of prospects. Pulling together highly specialised functions that can be operated in a fully secure and audited manner that supports these specific processes can be far more effective than trying to put in place rules and exceptions for a monolithic application.
However, a composite app requires other aspects to be taken into account as well. For example, it may be that if you are running your private cloud in a co-location data centre, there could be others in that facility that could offer you some of those functions you require at data centre speed and security levels, bypassing the need to go to an external provider. Co-lo providers such as Equinix and Interxion offer the capabilities for customers to share functions and services in this manner.
Others, such as Dell, are aiming to be cloud aggregators or brokers. They will provide their skills in pulling together cloud services and functions from trusted partners to allow composite app to be built and run in an audited manner.
Cisco is aiming to play at being both a provider and an aggregator: its OpenStack-based play is available for its partners to white label or to run in their own data centres, with providing the wherewithal to enable integration and audit.
The one fly in the ointment currently is VMware. It has decided to go its own way with vCloud. It is providing a comprehensive set of management tools around the platform and is making a play for public clouds so as to be able to play in the hybrid cloud space.
With its strong presence in commercial organisations, it will be tempting for companies to choose vCloud as their private cloud and to look for external public clouds that are either based on vCloud as well or are strongly compatible. If vCloud does ramp up and become successful, Quocirca believes that it will gain the same level of OpenStack interoperability that AWS and others have.
This move toward better standardisation whilst still managing flexibility has to be applauded. What once looked like being a Wild West approach to cloud computing has evolved rapidly to be a more sensible one.
However, we still have to see if this sanity will prevail in the longer term. If all implementations of OpenStack were the same, it becomes a drive to lowest cost. While some may try to differentiate based on cost, some will try to differentiate on functionality.
As long as this is layered on top of OpenStack and not embedded in it through changes to the kernel, and as long as this does not make the version of OpenStack incompatible with others, all should be OK.
The big loser in all of this would seem to be CloudStack. Although it gained more users in the early stages due to its greater maturity compared with OpenStack, OpenStack has gained the greater support from the technology vendors, and seems to be becoming the platform of choice for cloud providers.
It is just unfortunate that history tends to paint a picture of standards becoming fragmented over time. For cloud to really provide on its promise, a strong level of fidelity is required across the platforms. Let’s try and keep the momentum going in the right direction: choose clouds that try to maintain a good level of standardisation.