How to Avoid Cloud Vendor Lock-In
Vendor lock-in occurs when an organization becomes overly reliant on a single vendor for too many solutions and/or services, noted Chris Smith, chief marketing officer at Zenoss. "If you are relying on one of these companies for all of your infrastructure, it becomes very painful to ditch any of these items and move to a different vendor," Smith said.
Nov 13, 2013 5:00 AM PT
Vendor lock-in has been such a standard part of enterprise IT over the years that it often goes unnoticed and unquestioned. Recently, however, that lock-in mentality has followed enterprises to the clouds.
One might not think that vendor lock-in would exist for those who use open source software or open cloud solutions. Think again: Unless you take precautions before you plug into any cloud provider, you more than likely will experience one or more layers of lock-in limitation.
"People think that open source is the de facto standard," Pete Johnson, cloud evangelist for cloud computing provider ProfitBricks, told LinuxInsider. "But in the cloud world, there are not a lot of open standards. This becomes a big concern when you get involved in using one service provider's derivative services."
Enterprises need to put their heads in the cloud before plopping their critical enterprise data there -- otherwise, what goes in may still be very costly to migrate from one cloud system to another. It is possible, however, to tap into cloud technology and leave with your data intact -- if you avoid the pitfalls that let your vendor lock you in.
More Dollars Than Sense
Whenever you talk about vendor lock-in, you are talking about your migration costs if ever you needed to walk away from your current service provider. The reasons for doing so could include technology, mergers or corporate purchases, explained Johnson.
Often, it is not a technology barrier that creates a sense of entrapment for cloud users; rather, it is the related financial cost to "hop" clouds, he noted.
"Vendor lock-in exists when the cost of switching from one vendor's technology platform to another is so material that the customer is effectively unable to migrate from that vendor's offerings," John Pierce, vice president of information systems for Tripwire, told LinuxInsider.
That results in a huge reduction in the benefit otherwise realized by switching cloud vendors, such as reduced prices or technology improvements. In some cases, the extraction cost can even eliminate any benefits of moving.
Creating Vendor Lock-In
Vendor lock-in occurs when an organization becomes overly reliant on a single vendor for too many solutions and/or services, noted Chris Smith, chief marketing officer at Zenoss. It's a situation that's often unavoidable with full-stack mega-vendors.
"If you are relying on one of these companies for all of your infrastructure, it becomes very painful to ditch any of these items and move to a different vendor," Smith told LinuxInsider. "The pain and costs of moving to a different vendor can be significant, and folks might stick to a substandard system to avoid major costs."
Cloud services providers try to achieve customer dependency through a combination of hardware, software and know-how, added Elco van Beek, chief executive officer at Jitscale. Lock-in also is an indirect result of the cloud customer's wish to standardize on a limited range of technologies and/or suppliers.
Why Is It a Problem?
"Lock-in can be a problem as it creates monopolies for suppliers with certain customers and as such limits the pressure to innovate," van Beek told LinuxInsider.
A second problem occurs from the way a cloud provider tweaks its infrastructure, according to Udi Keidar, vice president of cloud operations at ClickSoftware.
Global cloud providers often provide private application programming interfaces to enable companies to use their value-added services.
"When starting to use these APIs within your code, the vendor actually limits its ability to move to another cloud provider since they might not use the same services or API, Keidar explained. "This, in a nutshell, is the main cause for cloud lock-in."
Cloud Lock-In Less Foggy
Cloud lock-in is a topic that Adam Stern, founder and CEO of Infinitely Virtual, no longer talks about because the issue has already faded, he said.
"There is some lock-in, but in reality it is not as sticky as you think because technically unsavvy people can not pull off leaving without help anyway, and vendors are trying to make it as easy as possible for you to bring your servers to them," Stern told LinuxInsider.
Cloud technologies have improved since the early days, when vendor lock-in was more difficult to avoid, he noted. The open cloud technologies vendors pursued since have made it easier for users to move their content someplace else.
While some lock-in still exists, better virtualization has all but eliminated the problem where operating systems and open source software are concerned, Stern added. For example, the hypervisor that makes virtualization work has grown to four or five proprietary as well as open source versions. That makes everybody's virtual servers that much more portable.
"As virtualization continues down the road, people will feel even freer to move their virtual machines around," said Stern.
Cloud vendor lock-in is not so much a big problem today as it is an evolving problem, according to Rick Sizemore, director of the cloud computing practice within outsourcing firm Alsbridge. Tools exist to migrate data back and forth among the three big server entities in the cloud: VMWare, OpenStack and Amazon.
"No technical issues at this level prevent you from moving data elsewhere," Sizemore told LinuxInsider, "but there could be contractual issues with your vendor that set time limits on when you can pull out.
"That is also mostly true on the traditional software side like databases and commodity services," Sizemore added.
The issue unfolds inside enterprises with the customized applications they develop. Add those to a platform as a service environment in the cloud, and the infrastructure becomes very proprietary.
"So, to take a customized software package out of one specialized platform and plunk it down someplace else, you just can not do until you rewrite the application," he explained.
Engineering Effort Needed
While it may not be considered true lock-in, anything that leverages an ecosystem that somebody else has makes it very difficult to extricate your data, Sizemore said. If you engineer your software so it allows the vendor's pieces to sit outside of it, then it becomes easier to move it.
It is not nearly as bad as people think, he noted, and that's is also the case with open source software.
"When you place it in the clouds using a vendor's software or platform services, you are going to get process lock-in," he said. "Once you build one of their processes into your software, it is going to be hard to change it."
Open Source Not the Culprit
Open source standards -- or the lack thereof -- do not always cause or eliminate cloud vendor lock-in. For example, from a technical perspective it will be really hard to move from a .Net platform to a Google platform without rewriting your code, explained Sizemore.
The situation gets even more sticky if you decide to embed a proprietary solution via open standards technology into other things that you do. Just because you build something with open source standards does not necessarily make the end result open.
"Where those two components do not mesh together, the lock-in sets in," Sizemore said.
Compounding the Problem
Knowing what your vendors are selling you is essential to avoiding lock-in problems in the cloud. In fact, many vendors sell proprietary solutions made with almost exclusively open source components, warned Sizemore.
It is the mixture of them together that makes it a proprietary solution, he explained.
Think in terms of basic chemistry, he cautioned: "There is a difference between a compound and a mixture. That difference is what causes the lock-in."
Software makers are well along on the path of reducing lock-in for cloud users. For example, a component that enables local virtual machine users to integrate their VMWare with the VMWare in the cloud makes it easier to move data between the cloud and the local infrastructure, suggested Stern.
It is also very conceivable that there will soon be APIs that will allow cloud customers to connect multiple cloud providers to their enterprise. This will let you literally move VMs from cloud to cloud with no barrier whatsoever.
"As we continue moving down the road we are on, lock-in is going to completely disappear," Stern predicted. "Users will be able to move their VMs around among different service providers very seamlessly."
'Lock-In Is Huge'
Stern added a provision, however. That seamless portability in the clouds is true as long as we are not talking about Software as a Service or Platform as a Service offerings; those technologies are a completely different operation.
"If you consider a company like Salesforce, then lock-in is huge," he explained. "That system is built around an entire CRM database. Once you build your data into that system, you have tremendous lock-in. Add to that the cloud, and you have a situation where you can not even consider the possibility of leaving."
Platform as a Service has very similar issues. In that case, once you build your software for a particular platform, you must start from scratch to move your software.
"That is part of the concept of building your software into a platform," he said. "Once you do that, there is no where else to go."
Enterprise cloud users should look for a software vendor that knows how to support the different cloud providers with its API. It should also use the services of a software vendor that provides an abstraction layer between your code and the cloud provider, suggested Keidar.
Another strategy to avoid vendor lock-in is to keep implementations as "plain vanilla" as possible. This will reduce costs and complexities of switching to another platform, Pierce added.
While it may be tempting to take everything you have and throw it into a public cloud like Amazon Web Services, you risk relying on one cloud for compute capacity. If AWS experiences downtime, your customers will come knocking. Your initial savings are unlikely to outweigh the cost of losing customers or downtime, noted Smith.
So, do not put all your eggs into one cloud basket. The savvy IT leader should take a hybrid cloud approach instead: House your mission-critical applications on your private cloud, where you have the control, but use AWS when you need quick infrastructure to build, according to Smith.
'Eyes Wide Open'
It is essential that companies approach cloud services with open eyes that look beyond compute and storage to help avoid lock-in, said van Beek. The strategy employed should be based as much as possible around platforms and technologies that are open and are based on market standards. Provider-specific technologies, platforms and software solutions should be avoided at all costs.
Keep your eyes open, ProfitBricks' Johnson strongly agreed. It is cheaper in the short run if you don't worry about lock-in, he surmised.
"When you realize you are locked in, then it becomes very expensive," he said. "Just go into it with your eyes wide open."
A Platform-Agnostic Layer
Service-level agreements, or SLAs, do not do much to help prevent lock-in, according to Johnson, so a key step in avoiding cloud vendor lock-in is to implement a layer that is cloud platform agnostic into your enterprise architecture and into your application landscape.
In this layer you can implement the application programming interfaces of various cloud-providers via a management dashboard based on requirements in terms of public, private and hybrid services. Then compare that against cost, performance, security and governance, said van Beek.
"An enterprise cloud user can certainly survive without buying into all of those derivative services," he concluded. "That is what happens with on-premises cloud storage."