Cloud - Does it really save costs?

One of the trends which everyone today talks about is Cloud Computing. Cloud is always projected as a "COST SAVING" initiative. However, does it really save costs in the long run?

No doubt "CLOUD COMPUTING" has changed the perspective of computing in today's world. It has its own unique place in each and every organization. Every organization has to look at the value proposition of the offerings in the cloud. It is not always about replacing your data centers or servers or storage with Cloud. Let us look at some of the specifics on where Cloud can help and where it does not justify to use cloud.

Let us start with 2 scenarios.

First is a startup which has its own unique value proposition and does not have any physical assets to begin with - think of the likes of the Digital Giants of today - Uber, Netflix, etc...

Second is an established financial organization which has already invested a lot in its own physical infrastructure and has a lot of established/working applications in its estate.

For the startup in our scenario 1, the company needs to focus the getting their business model ready and push their products/services to the end users as quick as possible. Being agile here is of utmost importance. Also, the company cannot firstly spend a load of their cash funds to establish a working data center let alone thinking about the disaster recovery scenarios. Assuming that they have loads of cash to burn for their infrastructure but time is also an important factor here. They cannot wait for the infrastructure to be ready to provision their services/products.

Let us take it a little further and assume that they have time also (which is very doubtful) but let us assume that as well. Since this is startup and we know the kind of quick adjustments that the technology stack has to go through to support the changing business demands, this startup also now cannot cope up with the traditional model where if a particular approach does not work and requires reworking, a lot of efforts have to be put in to realign the provisioned servers and applications. At this point the company also needs a strong infrastructure team to manage all these moving parts!!

As business keeps on bringing in changes, the focus and money starts to move from generating business value and unique products to managing the physical infrastructure estate of the company. At the same time, the company also has to spend a lot of time installing and configuring a lot software and related components to get see their products/services as the customers would see.

Now let's bring in the cloud here. The business team comes up with a requirement for new prototype using a specific set of technologies. Rather than wait for the physical infrastructure to be ready, the team can quickly spin up the required resources in the cloud, do their prototyping within minutes/hours rather than weeks/months. If they do not like the direction the product/service is going into, they simply shutdown/discard the resources and stop paying anything for them. Now the business model is truly agile where-in the business drivers can expect the resources to be made available to them in minutes and the charges are only for the time they use those resources.

Pretty neat. A lot of savings and a much faster turnaround without any initial investments. What more could a startup need at the early stages of their lifecycle. They have access to virtually unlimited resources within minutes. A key foundation to delivering a business service in an agile manner.

But wait, there is a lot more. The clouds of today not just provide compute resources, they have already a huge catalog of ready to use services ranging from a simple service like converting text to speech to more complex machine learning/artificial intelligence models. All at a click of a button - readily accessible within minutes without paying up huge loads of money. Now the startups can tinker their products/services at a much rapid and more so at a much higher degree of openness as they are not bound to any specific technology stack/solution. They can change any component at any point in time without worrying about their investments as most of the resources have not required any upfront investments. All services are in a "Pay for what you use" model.

All in all, the cloud is a big savior in this scenario and has helped numerous startups to do what they could have never imagined had the cloud not existed. So "CLOUD COMPUTING" has not only saved these startups from making huge investments but they have also protected these organizations from the potential pitfalls of worrying about making wrong choices and losing money in investments in physical / software stacks.

Now let us look at the other side of the spectrum and move to our scenario number 2.

The company already has an established data center and probably has enough compute resources to prototype small changes to a very limited scale and technology stack. However, this is more or less the actual scenario where as a standard you do not want to greatly deviate from what you already have in the estate. But still, such actions do seem to generate more value if cloud is considered here and exploratary work is not limited by what is already available in the estate.

Let's now look at the usual sales pitch that is done to such organizations - instead of reinvesting in hardware upgrades move the workload to the cloud. The key change which can deliver value over a long term here is often overlooked. This is the lifespan left of the data center which the organization owns. If the time has come to build a new data center the cloud move can be a potential saving however this is not always the case. The data center asset is not something which companies build every few years. This asset is a pretty long-lasting asset - more like a real estate and not like the technical estate which needs refresh once in 5-7 years. So, if the investments in the data center are already made, the cloud option for regular day-to-day workload will never turn out to be profitable.

This is further compounded by the fact that more and more hardware vendors are now offering their solutions at a very low margins making it really difficult for the established companies to move to the cloud. The use case where a server has to run 24*7 with more or less a known workload will always run cheaper within the existing data center rather than running in the cloud. Having said that, it does offer the freedom from any hardware upgrades and related downtimes which must be done for on-premises hardware once every 5-7 years. But does the differential in savings warrant such a move? I personally do not think so.

So, what does the "CLOUD COMPUTING" offer for such established organizations??

The simple and short answer to this is innovative services. My advice here would be use the cloud for rapid prototyping and tear it down as soon as possible. Do not run extended duration workloads in Cloud especially the "Infrastructure as a Service". These will never provide the benefit of cost savings in the long run. Use the cloud to run a workload which has very high variability where for certain time you need to scale to unprecedented levels like training machine learning models with a huge data set. But once the model is trained, shutdown the resources till they are required again. Use the "CLOUD" to exploit the ready-to-use services like cognitive services, data streams/pumps/analytics engines. Once the high compute tasks are done, move back to the on-premise capacity to run the day-to-day workload with a known variability.

If you follow this approach, the "CLOUD" will definitely save you a lot of efforts and make you more agile and innovative generating a lot more savings and increasing profits than to just "MOVE TO CLOUD".