We all know cloud and we all should (by now) know – why cloud. However, it would appear that some of the nuances that are part of IT decision making still escape many and in some cases are just ignored in the hopes that if you yell loud enough you can overcome any public hesitation to adopt. Cloud adoption isn’t a straight line and not just for technical reasons.
CIOs and Cloud providers do the industry a disservice when they tout “We’re moving everything to cloud” or “You don’t need a data center and infrastructure anymore”, context is king here. There has never been a time in IT when a single answer was all you needed, why do we think a single answer called cloud would be different. This isn’t about whether cloud is good or private infrastructure is better, this is about the realities facing most businesses when it comes to all or nothing approaches to IT.
It’s all about the Benjamins Baby
I (and many others) argued loudly starting a decade ago, that you don’t adopt cloud for cost savings. The fact that cloud isn’t a destination for Benjamin hunting has now been proven, but it’s still not clear to many why in some cases the costs are harder to capture or savings harder to create. There are a number of reasons why you might not save using public cloud, but I’m only going to focus on two, staffing and legacy IT and the breakdown of IT applications in the average business. Reads more like three things, I know.
Staffing and Legacy IT
The majority of enterprises don’t have 40 people that just do sys admin work, nor do they have 20 people that only do rack and stack. Most businesses have, by virtue of necessity, staffing models that dictate multiple hats for many in the IT Ops organization. The multi-hat issue means that if you move 100 servers worth of applications into public cloud, you are unlikely to be able to realize a one to one savings on headcount. Unfortunately, you can’t eliminate or repurpose effectively, 1/4th of Mark, 1/3rd of Aakash, 1/5th of Angelo and 1/10th of Vicki. Each of these four individuals are also splitting their time on jobs like Active Directory or File Server support. So, while the above example doesn’t mean you can’t adopt cloud, it does mean that finding savings is harder and that getting to the promised land will take more than just moving applications, it will take serious architectural strategy. Strategy for people leadership and career development planning and strategy for how IT is designed.
The Complexity of an Enterprise Data Center
To the Cloud operator or industry pundit who hasn’t actually run an IT organization, I just want to let you know – of the hundreds to thousands of applications in an enterprise, only a small percentage are actually immediately appropriate as obvious wins for moving to public cloud. As a quick refresher, here are a few of the time worn favorites:
- Significant elasticity
- Geographic distribution (many locations required)
- Specific value add services (I.e., Analytics or AI)
- Rapid deployment cycles
- Remote access (works for most applications just as well from a colocation or enterprise DC)
- Etc., etc..
Unfortunately, the above benefits have nothing to do with up to 80% of the applications in an enterprise. Every application isn’t tier I, nor does it require either daily updates or infinite elasticity. The majority of applications in an enterprise can be updated once a quarter and can take a few days to deploy. The majority of applications don’t require more than 10-25% elasticity over the course of a year. While having speedy deployment and infinite elasticity are great, moving applications to public cloud that don’t benefit from those things will not bring you cost savings, they will in fact, bring you higher costs.
This is not an “anti-cloud” rant
I’m still a cloud believer, but I worry about the way we message solutions in the IT space “all or nothing” or “do it now and get on the cover of a magazine”. From my perspective, we are entering an amazing time when best fit solutions are almost everywhere, from application development to SaaS delivery to global cloud access and infrastructure as code. We have the ability to code against a dozen different chips now and we can deploy to the edge if low latancy of data volumes are an issue, the options are endless. The fact we have so many options doesn’t necessarily make IT’s job easier, but it does mean that if the answer is easy, there’s a high likelihood it’s not the “best” fit.
Not discussed but worth consideration
I see enterprise response to the pandemic as being enterprise specific. Each business needs to evaluate their risks and costs against the perceived additional protection having every application in the cloud might provide. Keep in might that people will still be needed, and additional work with the cloud provider will be required to provide you with actual disaster protection.
Digital Transformation & Edge Computing
It’s quite possible DX and Edge Computing will significantly increase the demand for new applications and distributed infrastructure. Many of these new applications will likely be served via a public cloud provider. However, keep in mind that as you increase your technology foundation and footprint you are doing at least two things; increasing overall IT costs, while also increasing the bottom and top line value of IT to the business. As technology becomes a bigger part of your business delivery and customer interactions, that increased value means there will be more pressure to run IT systems as cost effectively as possible.
This blog was inspired by a conversation on @Rob Hirschfeld’s #Cloud2030 meeting from 2/25. The suspects involved included @tcrawford, @rhm2k, @parkercloud and @LawrenceHecht