4 Reasons Why You Shouldn’t Try to Build Your Own Cloud

As an MSP you’ll always look for ways to increase profits and many MSP looking at the prices charged by Cloud services such as AWS, Azure and Google may be tempted to by-pass them and develop their own cloud service for their customers. This article explains why you shouldn’t.

• Cloud vs On Premise
• Reason 1 – Building a Cloud is Complicated Business
• Reason 2 – Planning for Failure at Scale Sucks
• Reason 3 – Does your Datacenter Reach Worldwide? Didn’t think so…
• Reason 4 – The Big Guys Just Have Some Things That You Don’t. Accept it
• What You Should Do Instead
• Wrap-Up

Cloud vs On Premise

It’s a cloud-first world these days, right? Many Managed Service Providers are jumping on the cloud hype-train, and with good reason! The benefits of cloud technologies are numerous and can truly be game changing for businesses and for the MSPs that work with them to manage IT. It seems like an inevitable thing that most services run by an MSP will eventually move from on premise to the cloud.

However, from an MSP’s perspective, the question comes down to the “how” of cloud services. How are you going to provide the benefits of “the cloud” to your customers? This often breaks down to who’s cloud are you going to use? Azure? AWS? Google? Many MSPs start looking into these platforms and are often surprised by the cost of consuming services on these publicly available platforms. This often sends them to the thought process of “Well, I’ll just build my own cloud”. I’m here today to tell you why that’s a bad idea. It’s a TERRIBLE idea, and I’m going to tell you why.

I get it. I really do. I’m a technical guy myself, and I’ll be the first person to try and engineer my way to some cost-savings, but trust me on this one. It may sound like a good idea, but it’s not, and I’m going to tell you a few reasons why and I’m going to compare building a cloud yourself, with some examples of the solutions provided by Microsoft Azure, simply because that is the public cloud offering that I’m most familiar with myself. There are several cloud hosting options available however the reasons why building your own cloud solution is a bad idea are the same regardless of which you prefer.

Reason 1 – Building a Cloud is Complicated Business

Cloud services can be wildly complex. You don’t need me to tell you that. The problem though comes when us technical guys see it as a challenge. I’m going to quote the famous Ian Malcolm from the movie Jurassic Park here in that, “You were so preoccupied with whether or not you could, you didn’t stop to think if you should”. The quote is very applicable here.

Sure, you could likely put together a highly redundant cluster that serves up VMs or hosted services to your customers. You could then likely put together some VPN solutions or MPLS configurations to connect them. You could make a highly capable storage mesh, and failover capabilities…etc…etc…  the list will go on and on, and I would bet money you’ll still miss something in your planning because I would bet money that I would miss something as well. Making your own cloud is that complicated and there are many bases to cover, but moving on…

Let’s then say that sure, maybe you got it right. Maybe you did cover your bases. What happens when you have to manage the platform on an ongoing basis? Let’s focus on one common management task as an example. What about a patch cycle? You have to patch the OS of the core components, hardware components, firmware, switch infrastructure, storage hardware…etc…etc. Who is testing those patches in an identical environment before rolling it out to production? Who is making sure all those patches work out together. Are you going to eat the man-hours needed to do that properly? Are you going to have someone standing by to install zero-day patches at the drop of a hat? If I’m a customer of your cloud services, my expectation is that you’re doing exactly, all of the above. What am I going to do if that’s not the case?

These are just some of the complexities of attempting to build and maintain your own cloud. It is by no means a comprehensive list, but you get the idea.  

Reason 2 – Planning for Failure at Scale Sucks 

What about when things break? Do you have a plan to failover all those workloads elsewhere in the event that’s needed? Do you have regularly scheduled DR drills? Do you provide your customers the ability to do this themselves?

Things are going to break. There is going to be downtime for single failure domains in your cloud for things like patching, hardware replacement, and unplanned issues. How do you protect those consuming your cloud from those types of operations? Does your solution to this problem failover the entire cloud, or just individual customers? Is it seamless?

If those seem like difficult questions to answer, that’s because they’re meant to be difficult questions. Your customers are depending on you to provide these services. Some of your customers may not be able to EVER tolerate an outage, so your cloud needs to be able to address these types of situations and/or provide the customer a mechanism in which to do it themselves. To me, that’s a bare-bones capability of a modern-day cloud.

For example, Microsoft provides a number of capabilities designed to address the availability needs of their customers. Availability groups in Azure address patching and hardware maintenance concerns. Azure’s cloud storage has the ability to keep copies of data in multiple regions. Microsoft provides you multiple mechanisms for dealing with failures and disasters in Azure, and in the event that they fail to do so for some reason, they have money-backed SLAs. Does your cloud do that?

Reason 3 – Does Your Datacenter Reach World-Wide? Didn’t Think So…

Speaking of availability, does your cloud footprint reach world-wide? Let’s say I’m a manufacturer that sells worldwide, would I want my website hosted in a single location? What if I want to host a web app that allows my customers to login and purchase from everywhere? Would I want that living on a web server in my own datacenter or distributed across multiple datacenters and multiple geographies to allow for natural disasters? 

Most MSPs I’m aware of don’t have that kind of footprint. Yes, there are some with multiple datacenters, but the fact is you can’t compete with the likes of Microsoft Azure and the others. As of the time of this writing, Microsoft Azure has 54 regions world-wide and is available in 140 countries. If you’re an MSP offering cloud services through a major provider like Microsoft, this would allow you to provide an unprecedented array of geographic offerings for your customers. That’s simply something you own datacenter(s) would not allow.   

*Above Image from azure.microsoft.com

Reason 4 – The Big Guys Just Have Some Things That You Don’t. Accept it

Now, I’ve mentioned the above reasons so far, and I’d like to point out that these are the more direct reasons for why you shouldn’t attempt to build your own cloud. However, there are many, many more. While the big players, like Microsoft, AWS, and Google have figured out all I’ve mentioned above, there is one final point I’d like to leave you with as you’re weighing this decision. The brutal fact is, they can do some stuff in their respective stacks that you can’t and won’t be able to in a self-made cloud. I could go into detail, but I think the list of services provided by Microsoft Azure will get the point across. Can your cloud do all of this?

What You Should Do Instead

I’m hoping you’ve arrived at the decision to not build your own cloud. I won’t hold it against you if you attempt it, but I’d like to think I saved you some pain and misery by writing this article. If you’ve decided (based on the above) that building a cloud isn’t really your thing, you’re likely wondering, what should you do instead? The answer really boils down to two options.

Resell a public cloud platform – This is likely the most obvious. Reselling a public cloud platform can not only be quite lucrative but can save your engineers (and customers) a lot of time and headaches! The big vendors all have reseller programs. Links below: 

• Azure Partner Program
• AWS Channel Reseller Program
• Google Cloud Partner Program

Host a pre-built cloud appliance in your datacenter – While this won’t address some of the geographical concerns mentioned above, it does give you something of a happy medium between building your own and selling someone else’s cloud. This option allows you to host a cloud that has already been vetted and built, and retain some control of the hardware. The only solution that really fits into this category (that I’m personally aware of) is Azure Stack.  

Azure Stack is basically all the power of Azure in your own datacenter. It comes in a pre-built appliance format and is partially maintained by Microsoft on your behalf. If you’re looking for more information on Azure Stack, you can find the official site here, or you can also view a webinar we hosted with Thomas Maurer specifically about Azure Stack Here.  

Wrap-Up

I could go on and on with more reasons, but the point is, building a cloud that is on par with the big guys is unrealistic. Sure, the public cloud may seem expensive, but the big cloud providers have figured out all the stuff mentioned above, and more. Not to mention they’ve figured it out at scale. So, if the main goal for your business is simply providing IT services to your customers, then why are you trying to build a cloud? Instead of building a cloud why don’t you serve your customers well and you’ll quickly hit a business ceiling even if you do manage to produce your own service. Focus on your core business of serving your customers and sell someone else’s cloud instead.

Agree? Disagree? Have comments? Have you made the attempt and failed/succeeded? Let us know in the comments below! 

Also, if you’re looking for more information about hosting solutions and MSP Tools in general. We’re hosting an upcoming webinar specifically about MSP tools. You can find more info and sign up here.