Lock-In should be a Big Concern

I’m convinced that customer concern about lock-in will remain one of the top four considerations in cloud purchase decisions into the foreseeable future. Along with reducing cost, scalability and improved performance, many companies are trying to lower the risk and cost of lock-in. The real problem though is that the thing that drives lock-in (all your eggs in a single cloud provider’s basket) also provides many benefits to both the customer and the cloud provider. 

With a focus on a single cloud vendor, the customer benefits from simplified management and seamless technical integration. The vendor benefits by increased per customer revenue and lifetime customer value (perfect lock-in results in a customer that spends every possible dollar with just one vendor and never leaves).

In the past I’ve promoted the benefits of single source cloud platforms and encouraged customers to tolerate if not embrace their friendly cloud overlords. As much as lock-in is to be avoided there are significant benefits to working with a single vendor that may outweigh the drawbacks. Let’s consider the unthinkable and ponder life with lock-in.

Having Options is Always Better

First off let’s agree that being locked-in to one vendor is not the best. You may prefer organic burritos at Chipotle over a Big Mac at McDonalds or wish that all fast food would just disappear entirely, but having options is always a good thing. A nationwide fast food monopoly of just one restaurant chain would inevitably lead to lower quality, poor service and higher prices. 

Just as it is better to have options with your fast food, so also it is good to have options with your cloud platform provider. But providing options is not always that easy. Minimizing lock-in incurs some short term costs for some often unrealized long term gains. Benefits that may not accrue to the person making the decision.

Short Tenures Result in Diminished Concerns for Lock-In

When considering lock-in let’s also recognize that not all corporate purchasing decisions are made in the best long term interest of the company. Often they are made in the best interest of the decision maker and then positioned to be beneficial to the company to build consensus (see Samuel Culbert’s book “Beyond Bullsh*t – Straight-Talk at Work”).

Since the half-life (tenure) of the average CIO is 4.3 years and just 2.4 years for a CDO, considering the long term effects of lock-in may not be a personal priority. If you’re not planning on sticking around for very long you may not care about the problems that lock-in causes five years down the road. Your company, however, should care and so should your CEO (average tenure = 8 years) but the CEO is often reliant on his or her direct reports for making these decisions.

For these reasons (and others) organizations often don’t have a solid long-term plan for their cloud strategy. Instead, the dominant drivers of cloud usage can be short-term tactical factors like:

• Technical convenience => “We’ll spin up our next app on the cloud that is already there.”

• Political convenience => “I can get this project off the ground without IT involvement if I use my existing cloud vendor.”

• Data gravity => “I’ll put my next 10TB next to my other data in the temporary cloud to minimize data movement.”

• Preferred vendor agreements => “AWS is already approved… I’m not going to do the paperwork for Azure right now.” 

• Need to show personal success => “I can get a single sourced cloud strategy up and running in half the time of one that addresses lock-in.”

With short tenures of key decision makers and many factors pushing towards the tactical rather than the strategic, it is no wonder that lock-in happens.

Managing Lock-In is Still Important

But lock-in is bad. There is no doubt about it. It should be tamed and managed as best as possible. Here are two reasons lock-in has risen to the top of the list of concerns for companies moving to the cloud:

1. Pricing Asymmetries of Data Movement. Today, Hadoop and other open source initiatives have made the cloud a reality. Open source has driven down price and increased scalability but now customers are asking for more complete and robust cloud platform solutions. But the cloud platform providers (Amazon, Microsoft, Google, Alibaba, IBM, Oracle and others) have a vested interest in making it hard to move from one cloud to the other (note: it is usually $0/GB to move your data onto a cloud platform and ~$0.10/GB to move it off the platform). This in-out pricing asymmetry is pretty much the definition of lock-in. It is similar to the hassle barrier you face when moving your online checking account to another bank.  Yes. You ‘can’ do it, but it will be more painful and expensive than it seems that it should be.

2. Costs Will Creep Upwards. Cost savings and scalability are the main reasons that Hadoop took off. Hadoop, and open source in general, allowed for the populist revolt against database vendors that were too expensive. But the big vendors are desperate to get back to milking that ‘complete solution’ cash cow and part of that will be increased pricing and complexity of features and solutions. This means that, like your cable bill or the landscaper cutting your lawn, costs will find a way to creep upwards if not carefully monitored.

Your Options for Mitigating Lock-In

There are no perfect solutions to avoiding lock-in since it arises from the conflict between your desire to form strong relationships with your vendors and the desire of those vendors to lock you into their platform. However, you do have some options (not all of them perfectly desirable):

1. Embrace Open Source. This may not be the best idea since it means you have to have serious internal resources to keep up with the latest and greatest releases. In general, Open Source is going to change more quickly than what is requested by large slow moving firms. It will be hard, for instance, to convince a technically conservative large manufacturing company to use open source. Their pace of change may be so slow that they are just now (2019) debating whether to move boldly into the future and leave Lotus Notes behind (yes it still exists) and embrace Microsoft Outlook or stick with it for one more year.  Launching the third new point release this month for your cloud infrastructure is not going to match the corporate culture at a company like this. But still, about half of hybrid-cloud users employ this approach.

2. Look to Middleware / Fabric Vendors. Although the use of the term ‘fabric’ seems to be on the wane, the idea of middleware that knits together disparate pieces of cloud technology is moving towards ascendancy. Look for solutions that make it easy to move between cloud infrastructures. Almost two-thirds of customers are choosing this approach according to Fujitsu. 

3. Pay the Price for Redundancy.  Some applications and industries may require you to have strong back up and redundancy. This is an opportunity to run your cloud in two different places. It’s a good idea and keeps each vendor honest but it more than doubles your cost and, of course, the more places you put your data (even super safe places) the more likely it is that you could have a breach. Avoiding lock-in is one of the main reasons we see multi-cloud (multiple cloud vendors) and hybrid cloud (private and public cloud) becoming so ubiquitous.

4. Accept Your Cloud Overlords. One viable strategy is to go all in with one cloud vendor. Negotiate favorable pricing and simplify buying and contract management. This allows you to focus on solutions, not infrastructure and on SLAs not SQL portability. One suggestion here is that, at the least, you should have regular data dumps to another platform. In a recent report from Fujitsu, roughly one-third of customers accepted vendor lock-in as inevitable.

Deep Dive: A Middleware Solution to Lock-in

Solving lock-in will be an opportunity for some vendors. They know that there is institutional memory at large customers who were previously locked in by the data warehouse vendors. To solve the lock-in problem they are building a middleware layer that will make it easy to move from one cloud platform vendor to another. But this will be challenging as different cloud vendors implement APIs and services slightly differently from each other.  It will be similar to 15 years ago when all the database vendors said they supported standard ‘SQL’ and then you realized that there were actually slightly different dialects like Oracle SQL and SQL Server SQL that weren’t quite compatible and not optimized in similar ways.

This tower of babel balkanization of ‘standards’ will occur with the cloud platform vendors as well. Performance will eventually drive the decision for CIOs and CDOs between optimal speed and portability. This decision will come down to: “I can run my system on any cloud or I can optimize it to get 2-10x performance improvements by customizing to one cloud vendor.” 

This is a bit of a déjà vu experience for me. In the 1990s Sun Microsystems assured me that they had a PC simulator on their UNIX boxes and that I didn’t need to buy PCs. The reality was slightly different. It turned out that I could run PC programs on my Sun workstation but it was so slow that is wasn’t worth doing. Today the situation different. Getting PC code to work on a UNIX box was just a marketing checkbox for Sun. These cloud fabric vendors are very serious about building a highly efficient middleware layer so expect the solution to be very functional.

Four Criteria to Guide Your Decision

When making a decision about how to structure your cloud you’ll want to keep in mind these factors:

1. Cost. Your costs may not be lower with multiple vendors competing for the same business. Costs will come from many areas besides technical costs. You should also consider training costs and costs from managing multiple cloud partners if you choose a multi-cloud approach.

2. Cultural Appetite for Change. Judge your industry and your company. If you haven’t yet updated your corporate PCs from Lotus Notes then you probably won’t want to move from one cloud vendor to another very often.

3. Need for Speed. If you are desperate for performance you will probably need to optimize for a particular cloud vendor and get to know their strengths, weaknesses and quirks. But if being at the razor’s edge of fastest technology isn’t required, then a multi-source cloud vendor strategy may be sufficient.

4. Value of Redundancy. Some companies and industries may value redundancy more than others. For instance, running a mirrored cloud may be a requirement for certain regulated industries or even if you just absolutely positively have to keep your online store functioning at all times. Multiple cloud vendors will be a requirement so lock-in will be mitigated.

Conclusion: Accept Your Overlords

I like the approach taken by vendors endeavoring to create a middle layer that makes it easier to move your applications between cloud platform offerings. It is a good compromise that allows you to focus on just one cloud vendor while providing some insurance against lock-in down the road. You will surely pay some price in terms of performance and management complexity but it may be worth it.

All that being said, accepting your overlords and getting in tight with just one cloud vendor will be the default strategy with human nature being what it is. Don’t be embarrassed. It may also be the best strategy. At the least, it can be a viable strategy in the short to mid-term. Life will be very good for a long time before a major reevaluation of your cloud platform vendor is required, and, given that the half-life of a CDO is 2.4 years, the person making this decision probably doesn’t have to worry about the reevaluation phase. In fact, a strategy of embracing a single cloud platform vendor and then reevaluating every five years, could ensure lots of work and full employment for CDOs well into the future.

Related articles: 

• Four Signs You Need A Data Science Platform

• Don't Underestimate the Power of Stupid Artificial Intelligence Algorithms

• Commodity AI and the Next Best Experiment

• Age and Tenure in the C-Suite: Korn Ferry Institute Study Reveals Trends by Title and Industry