Yesterday’s MSP In Today’s Cloud: Doing Us All a Tremendous Disservice

July 25, 2018

Moving workloads to the cloud is among the most important and challenging initiatives for many companies. Legacy MSPs that have a hardware-first mindset often fail to effectively operate in cloud environments – leading to negative customer outcomes and sometimes reverse migrations off the cloud.

The processes and operational methodologies that made sense in a conventional datacenter are generally unsuitable for cloud environments. In addition, legacy MSP tools, processes, and training are often set up with faulty underlying assumptions that do not fit the deceptively-different cloud environment. Either the MSP fully understands and embraces the new world or it is stuck in the past - at Cascadeo, we want to help you avoid this negative outcome, and offer a few reasons to help you understand why this is true.

  1. Legacy MSPs build and operate systems that assume stable infrastructure. This is the number one reason Cascadeo sees cloud migration projects go awry and is the most preventable. Investment for redundancy at the infrastructure layer requires enormous time and expense compared to redundancy at the application layer. Cloud environments, by contrast, often are designed to contain cost and maximize flexibility - but are not designed for perfect VM uptime or data storage durability, except in very specific cases.
  2. Legacy MSPs assume consistent operating conditions and human performance. There is a reason Google, Facebook, Amazon, and other major sites seem to never go down - it is not because they have perfected operations or software, but because they focus on self-healing, distributed systems that anticipate lower-layer infrastructure failure by design. This goes beyond "DevOps" — although that is a key part of it — and really has to do with making an engineering investment rather than accumulating more technical debt silently through inaction.
  3. Legacy MSPs assume subject matter experts will be able to recover from a variety of fault conditions. People tend to do what they know: what has made them successful, gained them recognition and promotion, earned them certifications and the respect of their peers. This is not inherently wrong, but it creates a strong aversion to operational processes which feel contrary to it. For example, the frequent replacement of a production application server is unusual in a legacy datacenter, but companies that are cloud-first typically do this as an ordinary course of business activity. Break-fix / incident response activity in yesterday's world largely centers around human-driven, reactive processes in response to an outage or impairment; cloud-native companies use telemetry, analytics, and automation to prevent most outages and mask failures from downstream users.

How to avoid these issues? Choose a cloud-first MSP who recognizes that these implicit assumptions are fundamentally flawed in today’s world. Simply moving workloads to the cloud, via “lift and shift” or other quick-and-dirty workload migration approaches, will yield poor outcomes because it passes over the golden opportunity to refactor and automate - while promoting the same broken operational processes that we’ve struggled with for decades.

In reality, the cost of a “lift and shift” is often the same (or greater) than doing a migration correctly the first time around. The notion that a quick and dirty migration is better, faster, or cheaper has been thoroughly disproven by real world customer experience and market data. Cascadeo strongly advocates for an automation-first approach to cloud migrations and works to educate our customers about the many reasons why this is necessary as part of our engagement model. In fact, Cascadeo will generally refuse to do “lift and shift” unless we believe that it is in the best interests of our client and have fully educated them of the risk and technical debt that they assume by kicking the can down the road in this manner.

In our next blog post on this subject, we'll arm you with a list of questions to ask potential MSPs. This will help you evaluate whether you’re talking to a legacy MSP or one who is truly cloud-first and can help you to be successful on an ongoing basis.