Cloud repatriation is not the right way to look at it, understand the cloud computing continuum

There is generally conversation around cloud repatriation and there are lot written about on this topic.

In my view, to discuss on this topic, need to look at the point of departure for those enterprises, when they took an infrastructure centric migration to cloud.

1. The hardware was ageing and needs to be refreshed, and then hardware for the data center involved capital expenditure.
2. The data center did not meet the modern standards and most of the data centers were running in own premises whose tiering was low. Also, the data center was running in prime location like in the corporate office, where the real estate was very expensive.
3. When running in co-location facilities, the lease was coming up for renewal and there was a need to provide another long-term commitment in the renewals
4. Getting the required networking into the data center which includes backhaul, internet and WAN, involved lot of work in the networking side.
5. Skills to run the data center facilities was expensive compared to the size of the data centers
6. Green computing standards were emerging, and they could not be addressed in the existing data centers.

?

So 3 to 5 years back, the business case for accelerated migration to cloud in an infrastructure centric manner looks attractive. As the enterprises moved to cloud, the realization of the business case was challenging, there are few reasons for the same.

1. Optimization of the estate assumed at the time of business case, once migrated to cloud like usage of latest processors, etc. did not materialize fully.
2. There were many unknown costs in cloud which could not be estimated like network cost, monitoring & management tools cost and more. [These are more than 10% of the total cost.] The storage cost also tends to increase to meet the performance needs and also data growth, as business applications start collecting more data now, as part of the digital transformation.
3. Uncontrolled usage of cloud resources for new project initiatives.
4. Usage of the data center era tools on the cloud to manage, and they do not provide required feature set and do not take automation first approach. In the cloud everything is software defined, so pervasive automation is required to operate efficiently
5. Cloud FinOps engagement brings in certain level of operational discipline and cuts down the wastage in 1st few months once established. But cannot drive down beyond a point.
6. Hyperscalers stopped the incentives provided earlier and also the discount levels offered at initial stages vanishes, further negatively affecting the business case.

Is it all that bad for the enterprises? Definitely not if they are willing to change their approach of adopting cloud and other technologies. There are many old ways that are not any more relevant in the digital native ways,

1. Hybrid cloud is a necessity and treat the cloud as cloud computing continuum, and there would be a need to move across the continuum like public to private, private to public, public to public, centralized to edge and more on a regular basis.
2. Like any digital native company, the enterprises need to invest for technology debt reduction on a continuous basis which includes moving the workloads across the cloud computing continuum as the situation evolves. [Best of the organizations have 25% in each sprint allocated for technology debt reduction right from cloud deployment zone, to architecture, to platform technologies used, to development and deployment tools & processes]
3. The private cloud is viable again,

• as the hosting is done only in large co-location facilities which are tier-4 data centers [on par with hyperscalers], who get green energy and required network connectivity.
• as the hardware is available in Opex model from the likes of Dell, HPE, Cisco and others
• as the reliability is extremely high
• as the toolset available to run the private cloud is also 100% software defined
• as the technology needed to run modern applications are available to run in the private cloud, even as managed service from the technology vendor
• as solutions are available from hyperscalers to run in own data centers to meet certain needs of compliance, data sovereignty and more.

4. To be effective with public cloud, its capabilities and capacities need to be appropriately used. Elastic, scalable, on-demand and more, using public cloud as a data center and using just IaaS features may not give all the benefit of public cloud.

So, it is not that cloud migration done in the past is an incorrect decision and cloud repatriation has to be done. So being right on the cloud computing continuum is essential and operate with the right approach.