Why public cloud?

In my work for Hitachi I meet a lot of different people with all kinds of roles. Technicians, procurement, IT strategy, architecture, CIO, etc. All have to align to their employers cloud strategy in some way.

I work with IT-sales and mostly it's storage hardware as a disclaimer and I'm quite far away from cloud in many ways one would say.

I do read, learn and discuss cloud a lot with customers and find all kinds of statements regarding cloud. In many cases the cloud strategy to me lacks the real reasons why you have a "cloud first" strategy. You typically don't move applications to the cloud to keep IT-infrastructure cost down, it's usually more expensive if you already have kit in your datacenter. Moving all of your virtual workload to public cloud could be more than 3 times more expensive. The move is happening anyway.

My hypothesis on cloud strategies are that organisations plan for the cloud for other reasons.

In a short "click bait" format the reasons are:

• The infrastructure aren't able to keep up with the demand
• Purchasing infrastructure is a big impact in the financials every 3-5 years
• Digital transformation projects "require cloud"
• Everybody else is doing it

Let's deep dive each of these statements in detail to figure out the real reasons why the cloud is an attractive solution for a future IT-platform.

Infrastructure aren't able to keep up with demand

The current infrastructure usually consists of a virtualisation platform, storage box and some physicals servers running the VM's and bare-metal servers. With that you have a network with security features like Vlan, firewalls and a lot of routing. Adding to its complexity you run a SAN infrastructure to deliver the storage capacity and performance to your platform. It's a fairly static platform and it takes quite a lot of staff, complex internal process to order new capacity and with a wide range of SLA's on delivery between days and months.

All of this will give you a reliable solution including standardised data protection policy but not the fast moving agile infrastructure you see in all the demos from cloud providers. You have probably been involved in various automation projects aimed to cut the lead-times down and integration with solutions like ServiceNow or similar. Those projects usually ends with some workflows for ordering a standard VM on a given network but the total automation rate of the infrastructure will hardly be more than 20%.

The attractive solution would be to move all into the cloud where everything looks to be automated and you don't need to expand your IT team with new staff members while you get a really nice looking portal for ordering new IT capacity. All this including the quite common promise of not having to expand your IT-staff.

Statements like this aren't really holding up since you can get a really well performing solution if you outsource your IT to an outsourcer like CGI or HCL and they run it all with defined processes and physical IT but with a high automation and utilisation rate to be competitive the price per capacity. You can see it as a optimised way of delivering IT in the way you already do.

Purchasing infrastructure is a big impact on the financials every 3-5 years

Yes, IT-infrastructure cost a lot of money, there's software and hardware combined that are big investments and the procurement process is also expensive in many cases and could take months to finalise. The following projects of implementation and migration of your applications to the new platform that can take months to complete and running two platforms at the same time comes with a cost.

Purchasing large storage boxes are big investments and the IT-growth per year is usually about 30% in storage only in many larger enterprises, that would consume all of the price erosion you typically see per year enabled by new technology. Many vendors also have smart people working with licenses modells to keep the revenue up at the same time.

The future investments should be made in a cloud-like financial model where you pay for the usage instead of pay for 5 years on day one. If you would be able to utilise an elastic model for your on-premise infrastructure you ought to have a 100% utilisation of your investment in IT every month, predictive cost, and a proactive delivery of the capacity required by growth. Most of this consumptions models also include SLA's on availability, delivery and could include other types of services ontop of "keeping the lights on".

Digital transformation projects "require cloud"

I see from time to time organisations and especially larger ones seeing the transformation of the whole company is only going to be achieved by moving applications to a cloud platform. This is a theory that's aided by a modernisation of core applications migrated to container platforms or other "something-as a service". It's true in many cases that you can be more digital by purchasing a cloud solution for a specific use-case such as financial applications for book-keeping etc. Still that aren't really a digitalisation of the core of your solutions with a low cloud adoption.

Many banks and government services have digitalised the paper process and it's a great start but modernising your legacy applications from mainframe to micro-services and containers will hardly accelerate your core business on it's own. A change in the way you run an application itself would most likely enable a lower cost of hosting the application while speeding up the pace of the development of the application.

Will it run better in a public cloud ? The answer is not that easy.

If you have a large amount of end-users interacting with your application and you have peaks and large portions of the months there is a low utilization then an elastic model would be beneficial. If it's a back-end solution with a fairly static load you would not get an operational benefit in terms of total cost of ownership or similar by running the solution in a public cloud.

In the end, you need to have the option to make informed decision on where to run your application, it could be to burst in a public cloud but most of the time run in your own platform outside of peak usage to make sure you have a high utilisation of your own hardware.

Everybody else is doing it

This is one my favourite topics in the IT industry. The trends come and go in IT and hypes get accelerated by companies like Hitachi, Dell, HP, IBM, Intel. Gartner, IDC and others. How to deliver any applications have been more or less constant over the last 20 years but the platform on where you run the application have changed over the years. It's converged, hyper-converged, private, hybrid, public -cloud. Is there anyone who even remembers the serverless plattforms?

Lately containers are the most common trends broken out from the cloud trends and I believe the reason behind that are the complexity of moving IT into a cloud delivery model. I see that OpenShift and other solutions are getting a lot of traction by corporates but right now the adoption at scale is fairly small but with a high level of automation with a nice portal for deployment.

To be fair when asking about the cloud strategy the response is not that everybody else is doing it but reading between the lines you see an anxiety on some parts of cloud strategies and it's a great project to have in the corporate strategy and in the yearly and quarterly targets to your shareholders.

Would you invest in a company that does not have a strategy to handle the legacy IT platforms?

Starting with a container architecture and deploying new applications in those platforms are a good start for a cloud strategy since it's the core component of an efficient cloud delivery, it will also give you the final decision on where to place that workload and when. If you can run your application in an on-premise container platform you ought to have a low level of vendor lock-in when it comes to the large public clouds.

Summary

To summarise on hypothesis on each of the resons stated I can conclude that many organisations ought to see the requirements rather than the features.

I talk to CIO's who say that they are moving to the cloud even if it's more expensive that current model and the rationale behind those thoughts is what I'm interested in.

I also hear a lot of references to successful companies like google, Facebook, Uber and others that are 100% cloud based. These are some really fantastic companies but a bit that's usually missed when you compare a bank or an insurance company with Uber you'll notice a large difference. Those companies have only ONE or TWO applications in total, their focus are on that application only and the infrastructure and architecture around that application is solely designed and built for that; like Facebook or AirBnB. Facebook was founded about 16 years ago and VMware was founded in 98 and got traction around the same time Facebook got founded.

In my conclusion of I can see that many strategies should look for abilities in their future IT solutions rather than setting out a cloud first strategy. I'll just assume security in the cloud is equal to an on-premise solution and also ignore European issues with legislations like cloud-act.

A strategy on cloud should consist of a strategy that requires IT-infrastructure to be elastic in terms of capacity but also in it's financial model, Scalable far beyond the foreseeable future, SLA's (including financial penalties if not met) while being automated for both deploy, changes and decommission.

If you are able to deploy such a solution and there are many out there, you have put yourself in the drive seat of the future where you can decide what to run, when to run and where to run at the time more suitable for your needs and requirement but with the option to keep all of it on-premise if you opt-out of public cloud for various reasons.

Thank you for reading and have an awesome day!