A summary point about vendor lock-in concerns: In most
respects, moving to public cloud does not represent signifi-
cantly more risk of vendor lock-in than most enterprises
already experience in their current on-premises or colo
environments. It is typical for a large enterprise to minimize
the number of vendors in order to simplify their manage-
ment. With the vendors they already rely on, the need to
move away from any of them represents significant friction.
This again reinforces the earlier point that ease of switching
vendors is just one of multiple, already baked-in factors that
influence any enterprise’s current and future technology
choices.
There is, however, one aspect of
cloud vendor lock-in that should
be acknowledged: your data’s
“center of gravity.” Moving data
into the major public clouds is well
supported via multiple means.
Moving it out is not tremendously
difficult, unless data volumes are
very high. But it can get expensive.
It is not an accident that inbound
data transfers are free, while out-
bound transfers are metered.
While the number of cases where
enterprises are choosing to take
out all their data from public clouds
is small, it is important that this risk
is understood in the context of
other decision criteria.
Myth #3: Choose open-source
technologies to avoid vendor lock-in
This one usually comes up when proprietary software is
compared to open-source software. The issue is that a pro-
prietary commercial software vendor can enforce predatory
licensing models, whereas this risk is much smaller with
open-source solutions. Again, there is some validity to this
concern based on certain past experiences, especially when
the solution combines hardware and software (ahem, Ora-
cle). Using free OS and application middleware helps mini-
mize the threat of those predatory
tactics. However, it is important to
remember that changing applica-
tion architectures represents sig-
nificant friction regardless of which
licensing model is used. In other
words, while open-source software
potentially lessens the size of the
club you can get hit with while your
feet are locked in the mud, you are
stuck in the mud either way!
In most respects,
moving to the public
cloud does not repre-
sent significantly
more risk of vendor
lock-in than most
enterprises already
experience in their
current on-premises
or colo environments
However, even this data gravity question further points out
the futility of over-emphasizing the idea that public cloud is
the harbinger of vendor lock-in doom. Look at it this way:
• It’s challenging to move data because of its gravity
• Applications depend on close proximity to data (for
minimizing latency)
• Therefore, applications themselves already enforce a
form of location lock-in that has nothing to do with
public cloud per se.
Is it really sensible to ascribe to public cloud a huge assumed
portability penalty when application proximity to data
already creates such significant friction?
Myth #4: Portability
and abstraction tech-
nologies solve vendor
lock-in
This idea is a continuation of the
trend toward abstraction that
began with virtualization nearly
two decades ago. Virtualization does bring some portabil-
ity, allowing machine images to run wherever the respective
technologies, such as VMware, Hyper-V or Xen, are sup-
ported. The next level of abstraction, containers, continues
this trend. Containerizing applications does offer some
freedom in terms of where applications run (on-premises,
private clouds, public clouds), and in that sense, it helps
lessen vendor lock-in. But it is necessary to point out that
containerization is not always an optimal solution for appli-
cation workloads.
Perhaps more importantly, even when containers are used,
there are many other layers to consider in an application
infrastructure environment, such as database technologies,
WINTER 2019 | THE DOPPLER | 77