Cloud vendor lock-in - problems and solutions
Computing Delta research looks organisation's strategies for making data and applications more portable
Silos are the sworn enemies of enterprise technology. Silos prevent information from flowing where it needs to go; silos mean duplicated effort; silos create inefficiencies.
In recent years, one of the most powerful weapons for breaking down silos has been cloud computing. With its umbrella-like reach, cloud provides a very convenient tool for joining together various disparate groups, systems and data stores.
Inevitably though, the powerful tool for breaking down silos will eventually become a one itself. This is the impetus behind the multi-cloud pick-and-choose approach.
Base: 180 IT leaders at organisations with 100 to 10,000+ employees
In a Computing Delta study we asked 180 IT leaders at organisations with between 100 and 10,000 employees that are using cloud IaaS/PaaS whether they see cloud vendor lock-in as a problem. Fifty-one per cent said they do, with eight per cent actively taking measures to avoid it, not a huge figure perhaps, but still significant.
The figure below shows the factors most likely to cause cloud vendor lock-in, according to our respondents. At the top, we have licensing and contractual issues, the cost of data migration, the existence of proprietary services or APIs and skills.
We looked at current cloud strategies within the organisations that we surveyed. This was covered in a previous article, but the important conclusion is that there's a desire for an integrated multi-cloud setup where data and applications are effectively cloud-agnostic.
See also Integrated multi-cloud is on its way up
So, what's the specific appeal of integrated multi-cloud? Well, the number one reason was the ability to choose best-in-breed services from different vendors. Related to that is the second issue: avoiding vendor lock-in. Other benefits perceived by our respondents were improved flexibility, superior business continuity and the ability to drive down costs through the power of being able to walk away.
Unfortunately, though, for most users an integrated multi-cloud remains a dream. We asked about the drawbacks in this approach to data and application management.
Deploying the IaaS/PaaS services of two or more cloud providers is very likely to increase management overhead, and this must be set against any benefits derived from the approach. It may also be necessary to hire more skilled professionals. While cloud IaaS and PaaS services are pretty similar there are important differences beween them and there's always a learning curve before teams can get up to speed on a new platform.
Then there's the need to management and monitor of workloads across different cloud platforms using tools that might not be completely cloud agnostic and integration costs must be taken into account. It's also possible that reseller partners will not support your cloud of choice.
For most, perhaps, the benefits of an integrated multi-cloud are currently outweighed by these issues, but it's certainly something that more organisations are looking are looking at.
"Done correctly, it would provide businesses with the ultimate in flexibility and cost savings," said one IT leader. "But we're a long way from this."
Seamless data portability requires modern infrastructure and this can be another blocker. We asked, "does your storage infrastructure get in the way of data portability?" and 80 per cent said ‘yes'.
The biggest problem, according to our respondents was the presence of legacy interfaces such as SAN and NAS. Then there's the unavoidable fact that data volumes are very large and growing all the time. And that's before you get to the difficulty in managing state across platforms that we touched on earlier, proprietary databases and a number of performance issues which can arise when local infrastructure is combined with cloud
There are a number of ways that organisations are trying to get around some of these difficulties to make their data more portable. Among these are hyper converged infrastructure and software-defined storage, and also the use of containers and Kubernetes which can make applications and data much less platform dependent.
Hyperconverged infrastructure/SDS was in use by 15 per cent of our respondents were while 16 percent were using Kubernetes and a further three and five per cent respectively were about to start adopting those technologies.
Multi-cloud is part of a logical progression towards greater portability of data and applications and CSPs are starting moving to accommodate it. Even AWS, which held out against the inevitable for quite some time, announced some tentative multi-cloud measures at Invent in December.
Indeed it would be strange if this were not the direction of travel, as it would mean that technology would have given up its age-old battle against the silo.
Delta is Computing's market intelligence platform covering key areas of enterprise technology and helping IT leaders to make the right strategic decisions.