security and identity management. Other abstraction layer technologies, such as Pivotal
Cloud Foundry, are a slightly different take on portability, enabling cross-cloud capabil-
ities. However, relying on a company such as Pivotal to minimize cloud vendor lock-in
creates lock-in with Pivotal. The takeaway is that abstraction technologies can indeed
play a role in minimizing dependence on a particular vendor in some contexts, but can-
not fully remove vendor lock-in, and may in fact contribute to it. Like every technology
or solution, the ability to change vendors with little friction is only one of many
concerns
Myth #5: Multicloud architectures are a great way to avoid
vendor lock-in
The rationale here is this: “I will avoid cloud vendor lock-in by using multiple public
clouds.” There are ample reasons why this is a less than optimal philosophy.
1. Every public cloud vendor is different. While there are many broad similarities
between the leading public cloud vendors (e.g., they all offer VMs, block storage,
Hadoop services, etc.), they are in no way identical. This means that the theoreti-
cal ability to easily shift workloads from one vendor to another is largely a fan-
tasy. While it is certainly possible to migrate many workloads, there is a huge
number of feature and implementation differences between the primary clouds.
2. Each cloud vendor has unique, powerful features. Public cloud offers some
attractive features that are often specific to a single vendor. Echoing thoughts
from the section on Myth #1, sometimes the only way to escape vendor lock-in is
to avoid using some of the amazing capabilities that drew you to the public cloud
in the first place. Is the risk of friction in moving from one cloud to another higher
than the risk of letting your competitors take advantage of capabilities and effi-
ciencies you have left on the table?
3. Multicloud adds complexity. Managing multiple cloud operating environments
increases the challenge of organizational change, staffing and management and
dramatically increases the risk of human error. It is already difficult enough to
train or hire cloud expertise, and managing more than one cloud multiplies that
challenge. This does not mean that using multiple clouds is necessarily a bad
idea, but pursuing a multicloud philosophy for the sole or even primary purpose
of escaping cloud vendor lock-in is almost never a wise idea.
A Smarter View of Vendor Lock-In
Although the perception around vendor lock-in assumes it is a bad thing that must
always be avoided, the reality is much more nuanced. Vendor lock-in is definitely real,
but the antiquated view about how important it is to avoid is based, as we have seen, on
a number of myths and misunderstandings.
78 | THE DOPPLER |
WINTER 2019