The road to Digital Transformation does not go through Hybrid Cloud

Digital transformation is a big word and it has the risk of being a vacuous word. There are signs that vendors are using this term to describe anything to do with electronics. I want to talk about digital transformation in a narrower sense. Lets talk about competing in your market using software. Think about improving customer experience so they keep coming back to you.

Digital transformation requires in-house software development. There is no turnkey vendor solution yet. You have to develop applications and invest in it. Status Quo is not a winning strategy.

It is likely that you already have several apps. Some of these can not be rewritten. Reasons for it could be shorter shelf life or lower priority or lack of resources. These are running in your data center. Your investment should have bias to new application development or refactoring existing applications.

The new applications need to be micro services based and should leverage elastic cloud infrastructure. These applications should cost you less in OPEX than previous era applications. For example, a well designed serverless app should cost you less.

The new applications can not run inside of your data centers. That would be a waste since DC imposes constraints and increases cost. These new applications can not also run a DC-like environment on the Cloud. Most people, excluding legacy vendor marketers, would agree with this.

The question now comes about what to do about legacy apps that do not make sense to refactor?

Should you forklift them to a simulated DC environment on the Cloud? Should you rent bare metal and recreate your DC environment there? Should you get an appliance like Engineered Cloud Machine or AzureStack?

The answer is no. Only thing this does is increase your costs. More dangerous, its an opportunity cost for you. Time and money is better spent refactoring your apps or developing new apps.

I have heard several arguments on why Hybrid Cloud makes sense. Let look at them:

1) Use same Cloud infrastructure as leaders

Crap packaged in a gift box is still crap. Just moving your existing legacy app to some bare metal in Cloud is not going to make it a cloud app. Your app won’t be more available either. If you add DR, it would be, but you can do that without moving the app to a bare metal on a Cloud.

For a fun exercise, take 8 powerful DELL Servers, cost of comparable instances in the Cloud. Your hardware requirements on Cloud will be like what they are on-premises, if app hasn’t changed. Yes, include expenses like power etc as well. Tell me if Cloud is cheaper than on-prem.

2) Low latency with rest of your Cloud apps

If you are integrating new workloads with legacy apps, you should rewrite. Without it, your primary concern won’t be latency, but the legacy app constraining the new apps.

3) Get rid of your Data Center

Yes, this is a valid argument. If you are working on real-estate optimization and/or power/space optimization, do this. But, in this case, look at not just cloud providers, but all colo vendors and select the one that is most cost effective. You may find that your local colo is cheaper.

4) Proven infrastructure

Yes, Cloud provider infrastructure has matured quite a bit and they know how to run large scale infra business than you do. The benefits are high when your app take advantage of the cloud provider services. A forklift approach will only leave you footing the bill for operational costs of the Cloud provider.

5) Cloud Bursting

Hey, have you tasted Unicorn meat? How does it taste?

In Summary, be cautious of forklift or “lift and shift” based recommendations for legacy apps. Yes, I understand each enterprise is different and your requirements may be different. All I am saying is that don’t buy into marketing and do the thinking for yourself.

Thoughts?