With many of our due diligence engagements, we often hear, proudly from the CEO, how they are “in the
cloud”, as if that is a special code word for quality and scalability — therefore, we don’t need
to
waste time looking any further, box checked. Oh if only it was that simple.
Fortunately, we take the ‘trust but verify’ attitude and perform a detailed look anyway, which is
good,
as we seldom find a setup that can’t be radically improved. The problem you see, cloud architecture is
hard — takes a little more than opening an account on Azure/AWS and spinning up some servers — one S3
bucket a cloud architecture does not make.
As the leader of your organization, how do you know if your technology team have done you a
(dis-)service? That is going to be hard for you to evaluate. Help is at hand though.
To make it a little easier, we’ve compiled a list of the Top 10 signs that may point towards an
under-used cloud opportunity. You can keep an ear open for these signs as you interact with your
engineering group.
If your cloud bill is roughly the same each month, particularly if you are a seasonal business with periods of high and low usage, then this points towards not utilizing the elastic nature of the cloud — paying for what you need when you need it.
If the engineering team schedules downtime with each and every release, it suggests they are using servers in a static mode, where software has to be updated on each server. Modern cloud practices encourages replacing of servers with each release (think containers, e.g. Docker), and switching over traffic when satisfied all is well, along with deleting older servers.
As the old saying goes, you can’t make a baby in a month with 9 mothers. In the cloud though you can and if you can’t introduce another mother to cope with load then your application is not cloud ready.
If you hear reports of servers running out of disk space, then this points to servers running in a static manner that requires manual maintenance. A good cloud application utilizes the effectively infinite, pay-as-you-go, storage.
No one wants to hear (or report) loss of data and it shouldn’t really happen in a good cloud environment (outside of intentional user deletion). This suggests single points of failure, with no real elasticity or redundancy to the infrastructure.
With the absence of hardware in your new virtual world, it is easy to assume hardware failures won’t affect you. While rare, they still occur, and if such a failure takes down a piece of your architecture and causes major disruption, then your cloud setup may be too tightly coupled with no redundancy.
Modern cloud providers offer a rich array of services that removes the need for you to manage individual servers yourself. For example load balancing, logging, caching, and configuration are all software layers that traditionally you would have ran up servers and managed yourself. If there are pieces of the architecture that fits into this, then ask why, as it may show a lack of knowledge or understanding of the available cloud facilities.
How you secure your traditional hardware based servers, is not the same as the cloud. If you have manually configured firewalls on each server, or have no real network separation then you may be missing out on a whole layer of security on offer from the cloud provider.
If you are still talking about DR as something you switch over to in the event of the day you hope never arrives, complete with a lot of redundant duplicate servers just waiting for said day, then you are definitely not cloud enabled. With a well architected cloud, DR comes as part of the equation free-of-charge.
Hardware is expensive, so it is common to run multiple services on a single server (think the classic Linux Apache MySQL Php stack). The same economic constraints do not exist in the cloud. In a cloud world, each server should be as light as possible, making it easy to replace, maintain and scale.
None of these signs, in isolation, should have any major cause for concern, but if you find yourself
recognizing more than a couple then there is a strong chance that the cloud isn’t delivering what you
think it is and for the cost it should be. When we do recognize such missed opportunities, and educate
the team, it is not uncommon to see their monthly cloud spend fall over 50% once we’ve finished. There
is a lot of wasted cloud resources being paid for that is not doing what you think it is.
Challenge your technology team. A good cloud infrastructure should reduce your overall IT monthly spend
and only change in line with traffic.
We have found that many executives are not fully equipped to know the signs or the questions to ask.
Likewise, many engineering teams that have evolved with the business are not fully aware of the
capabilities the cloud has to offer and how to re-architect to fully take advantage of said
capabilities.
It is a simple case of education on both sides. A situation that can be remedied. If you find yourself
on either side of the equation, then please reach
out to us to see if we can re-mediate the situation
and get you on the right cloud.