The end of the physical data centre ?

Mark Cutting Cloud, Consolidation, Contracts, Governance, Management, Planning, Strategy, Technology Leave a Comment

Yesterday I received a call from a former colleague and old friend, who (I think) was looking for some advice as to where I thought the IT industry was headed. On par with similar discussions on LinkedIn, it’s also a question I’m often asked in due diligence meetings. As it seems to be a very valid and pertinent question, below is a synopsis of my thoughts as to where this industry is headed.

With Microsoft announcing their cloud target of $20bn in October of last year (a year ahead of schedule), Amazon’s seemingly unstoppable dominance of the cloud market in terms of infrastructure as a service, and the ever evolving Google Cloud Platform, is this really the end of the road for the traditional data centre ? Nobody can realistically argue that they haven’t thought about or even adopted cloud services as part of their formal infrastructure strategy – and if they did, this wouldn’t be an entirely truthful response. Admittedly, there will always be those who give services like AWS, GCP, and Azure a wide berth, but this is primarily for compliance and legal reasons – with the secondary concern around security.

Understanding cloud offerings

Being the first major player, AWS has become easily the largest and most dominant provider of services within the sector. With an impressive array of service offerings ranging from a simple website to fully scalable and resilient technology solutions, it is the clear industry leader. However, as more firms begin to embrace the cloud technology on offer, and exclusively use SQL, Azure is showing signs of steady growth. What gives AWS the major advantage over Azure is the provision and support of the Oracle database platform, alongside its own Aurora database offering, which is something that Azure cannot (and probably will never) offer – Microsoft are keen to take custom away from Oracle and see it as a competitor – a threat. In this case, you can’t blame them for not providing this service, preferring to offer their own alternative products instead.

But how does this equate to an organisation that uses Oracle and SQL ? Whilst this sounds like a contradictory position, it’s not uncommon. A likely scenario is that software is developed in house against Oracle, but that same organisation uses off the shelf software solutions that require SQL. With hundreds of software vendors leveraging the free version of MSSQL or the open source platform of MySQL / MariaDB in order to keep their software costs low, you could argue a valid point in that the organisation should consider a move away from Oracle. If only it were as simple as that. Vendor driven applications typically support most common flavours of database formats, and Oracle isn’t an exception. They also (in most cases) provide a way to migrate to a different database format via a custom written utility or API. However, in house developed applications that are considered to be mature are always much more of a challenge – mainly because it’s not only the underlying database structure that needs to be migrated, but what could easily be millions of lines of code that would need review and regression testing before being considered stable under another database platform.

With this in mind, it is often the case that keeping Oracle as the underlying database is actually cheaper than a code rewrite – particularly as such an activity is labour and resource intensive arising from the need to fully test all aspects and functions of an application. Based on this somewhat obvious observation, organisations still need to reduce costs where they can in order to remain both profitable and competitive within their market sectors. A particular scenario could mean a historical initial investment in per CPU socket licenses for Oracle that ideally would be converted into a NUPS agreement in order to comply with any potential cloud offering. Oracle has always been very difficult to deal with when it came to licensing, and most organisations often found that in order to remain compliant, they would either have to license the product at physical host level, or go down the route of adopting the Oracle cloud service – at a premium cost. Oracle have released their grip on the market in this area, and have allowed AWS to provide Oracle effectively as a RDS based service – in this instance, you only have the database itself, and no access to the underlying operating system where your instance resides. AWS also make it possible for you to bring your own licenses, which also dramatically reduces the cost, and also means you do not need to write off the initial spend in order to be compliant at the licensing model. An important point to remember here is that you would need a valid subscription agreement with oracle in order to convert your licenses in the first place. Equally, if you do not have this in place or it has elapsed, Oracle typically penalise you for this and charge a premium to reinstate the maintenance agreement first before even honouring the upgrade (which also has a cost associated with it).

For organisations that do not leverage Oracle, the cost of database services from AWS and Azure are very competitive – and superior to what you can realistically run on premise. Hardware in this instance is virtual of course, and fully scalable. There’s a common phrase in the information technology sector which is very concise.

“There’s no money in tin”

This has been the ethos for a number of years now. Modern technology dictates a completely different ecosystem these days, and those companies that are trying to lock in their Return On Investment need to also consider the impact of their BCP/DR strategy (sorry about the “death by acronyms”), and the overall security of the product they are using. Moving to the cloud is usually cost neutral or perhaps negative in year one, but the real Return On Investment is the cost savings that can be attributed to other areas of technology over the coming years. There are also trade-ins for licensing with software assurance that reduce the overall cost.

Cost margins reducing for hardware

There’s little to no profit margin for hardware resellers when it comes to servers – with the cloud based data centre becoming more dominant than ever, and organisations looking to reduce their on premise infrastructure and associated footprint, the last thing any CTO wants these days is new servers to feed and water (unless they are a data centre themselves of course).

With all the increased activity around moving to a fully cloud based solution comes the inevitable question – survivability. In this sense, I’m alluding to the job market as a whole which in terms of 3rd and 4th line will probably shrink as institutions begin to claw back their spend on personnel previously required to support local infrastructure – by effectively making them redundant. If your estate is running in the cloud, the argument to keep on site support staff becomes harder to justify. First line support isn’t realistically effected for all the time a firm is using local machines and printers, but when they same firm begins to look at VDI and considers using their PC’s as dumb terminals to access a centrally hosted desktop, that could also change. As bizarre as it sounds, it is actually easier to justify desktop support rather than more senior staff – both from the cost and headcount perspective post migration to cloud.

The final comment I would make is that the days of companies using a small business server with a handful of connected clients are pretty much over these days – essentially, killed off by mobile working and the cloud. With a cloud environment, you can be anywhere – you only need an internet connection. Be that in the form of a local coffee shop, Regis arrangement, or your own home is the only real decision you’ll need to make. The rules, and by definition, boundaries created by companies with their own server farms have been broken, creating a wealth of flexible computing opportunities that firms of all sizes can leverage using a cost effective monthly pricing model. What is abundantly clear (to me, at least) is the inevitable impact this will have on the workforce. With AI and Quantum Computing gaining momentum on a daily basis, the clock is ticking and it cannot be stopped. How long will it be before the clear cost reductions in technology equate to less jobs for those in the affected positions ?

Any institution not willing to embrace the cloud will no doubt have their own regulatory needs, but small and large businesses alike can all benefit directly from it. Essentially, cloud isn’t just “another way” – it will soon become the “only way” to dramatically reduce costs and IT overhead as that is the core direction most industries as a whole are headed. 

Ask yourself this – will the companies you support or the firm your work for still insist they do not want to lose their Return On Investment with an aged server and software that they will have to pay for all over again to upgrade ?

About the Author
Mark Cutting

Mark Cutting

Facebook Twitter Google+

Mark Cutting is the founder of Phenomlab.com and Inocul8r.net. He is a network, security and infrastructure expert with more than 27 years service in the Information Technology sector. Mark has a significant eye for detail, coupled with an extensive skill set. Having worked in numerous industries including trading, finance, hedge funds, marketing, manufacturing and distribution, he has been exposed to a wide variety of environments and technologies alike.

Leave a Reply

avatar
  Subscribe  
Notify of