The trojan horse called cloud computing

Cataclysmically, the datacenter universe has shifted to cloud providers. While most enterprises are averse to using public clouds for their core offerings, realization is seeping in that the law of diminishing returns applies to on-premise datacenters. As applications are being built for scale, the stress of data gravity has turned foundations of on-premise datacenters into thixotropic clay!

Cloud computing is here to stay and that is a fact. Some prefer to think of it as the only way out for organizations. Others have a decidedly cynical view of the after effects of the cloud quake! The oft repeated quote goes “the benefits far outweigh the risks”. 

Let us look at the stranglehold that cloud providers are exerting on the market.

• For the first time, 2019 marks the year where enterprises are spending more on cloud offerings than for on-prem datacenters. In fact, their cloud spending growth of 56% is gigantic compared to a measly 4% for on-prem infrastructure
• Cloud provider margins are way higher than people anticipate. Currently their gross margins are as good or better than fortune 500 companies who have invested in specialized datacenters. Some estimates ratchet it up to almost a 100% on any service they offer. Sounds familiar? It is totally comparable to a true SaaS offering that commands 90-95% gross margins
• Data gravity is leaning the cloud storage way after a decade of continuous growth. While the on-premise storage market by 2023 is projected to be around $140 Billion USD, cloud storage is racing to $170 Billion USD by 2024
• Data gravity is also accelerating migration of workloads into the cloud and it is projected that by 2025 more than 3/4ths of enterprise workloads will be running in the cloud
• Enterprises are spending more on software to the tune of $457 Billion USD by the end of this year out of which cloud application market has shot up to 33%
• Cisco says that more than 90% of the workloads will run in the cloud by 2021 and the cloud provider market will quadruple

As we can see, the march to the cloud is inexorable and the potential is unlimited in the short term. Cloud providers are lining the yellow brick road with smart paving to ensure that the transition is seamless, accelerating and empowering. But there is a niggling fear of what this might mean given that the market has become a cartel shared by Microsoft, Google and Amazon. Cloud does not mean utopia, does it?

There is an internecine outcome of the cloud that drives business away from their core values and while some may call it agility, others would name it opportunism. Cloud providers are literally building the market the way they see fit and companies have to adapt or perish. What could possibly go wrong?

Job losses are inevitable

The obvious commentary is that re-skilling fixes everything. IT personnel can instantly become this baton twirling conductor who can manage clouds, code in python, design/architect the heck out of everything, DevOp the entire datacenter perched on their mobile app, erect business continuity frameworks, make budget plans, optimize running costs, carve out data strategies and morph from IT administrators to application owners. Even if that were somehow airlifted by the cloud, it would still be impossible for most individuals to multi task to that extent.

It is not surprising that average pay premiums for tech certifications (ITSCPI) are steadily in decline. It is also true that system architects are being preferred to system admins. Hardware technicians are no longer the force they used to be. Serverless computing is accelerating the cannibalization of pure programmers by full stack developers.

Test driven coding does away with integration testers, cloud automation kills system test engineers and DevOps leaves everyone else in their dust. A combination of Agile+cloud orchestration+SDDC erases specialized jobs like program management, QA/QE, performance assurance, etc. As this nifty little survey points out, jobs that are out there demand a minimum of 5-7 skills from people. A tall ask is the way these skills cut across hardware, software, applications, networks, infrastructure, cost optimization, etc.

Budgetary overruns are common

As most enterprises what they resent? It will be budgetary overruns compounded by inadequate tracking of application behavior. The heat is on every single cloud journey/project owner to make sure that budgets are predictable and overrun risks are minimized. Some of it is due to imprecise understanding of cloud dynamics and the other is because of non-alignment of transaction revenue to back end spending.

According to this RightScale report, the #1 problem to be tackled in the cloud is cost optimization and more than 27% waste is being seen in core projects. The problem is in instance sprawls, unused services, uncontrolled spin-up/spin-down of serverless computing elements, misconfigured orchestration, over compensation for workload performance, excessive migration/onboarding costs, unanticipated load and unrealistic expectations. 

Hardware competition is being stifled

This comes as no surprise and is a trend that has been seen for decades. Commodity hardware is killing specialized servers. Merchant silicon is ruling the roost in cloud datacenter networking. SDN’s primary outcome is to cut a broad swathe through the director class switch/router market.

The unpleasant side effect is that specialized hardware companies are driven to be niche, lower costs and deliver hyperconverged scale. By the end of this year, there may be almost 500 hyper-scale datacenters and most of these will be owned by a handful of companies. Imagine the buying power that forces hardware companies to supply products at lower price, run on razor thin margins and replace skilled jobs with robotic automation.

Cloud providers are sourcing hardware from large manufacturing firms like HPE or DELL. It makes no sense for them to manage hundreds of smaller vendors. All hardware innovation will now be centered around manufacturing volumes, FRUs, remote support, clustered computing, speed to market, etc. the point is that smaller server companies cannot scale up to meet these challenges and it is no longer profitable to run niche manufacturing assembly lines. The obvious stagnation of server technologies has resulted in hardware innovation focused around GPUs (or lookalikes nee TPUs), IoT, embedded devices, etc.

The largest beneficiaries of this “standardization” are ODMs that are starting to look alarmingly alike. Owning 47% of the market without focusing on large scale innovation tells us a lot about which way the server hardware market is growing.

Real innovation now belongs to cloud providers

Cloud companies are filing patents willy-nilly. The usual suspects (IBM, Microsoft, Google, Amazon…) have filed more than 5000 patient families (with derivatives). The arms race has shifted away from enterprises to the cloud provider and the cloud has become the crucible for any new-fangled idea. Startups are flocking in droves unwittingly allowing cloud providers to sniff out bleeding edge technologies that can easily become monetizable services in the future.

Cloud providers today account for more than 30% of the patents worldwide. The future looks bleak for software patents especially after Alice. Infrastructure patents should show corresponding declines as increasingly hardware gets deployed by cloud providers in their own data centers as well as part of edge solutions.

As SaaS services are consumed with regularity, umbrella patents keep getting filed and on-prem hardware drifts away; the obvious conclusion would be that patent filing becomes tedious, time consuming, infringement is assumed and the thirst for innovation slowly dies out

Bleeding edge business models are being thrust down our collective throats

If computing were a playing field, then all games being played are away ones. There is no more home-field advantage anymore. One needs to play by rules that are hard to adhere to unless the business is already contoured for it. Simply put, the whole world is available on subscription!

Does this mean that it is inexpensive? Far from it! I once worked with a client in the healthcare industry who wanted to archive his data in the cloud (all 150TB of it). It turned out that it was 10 times more expensive than on-prem storage even if the AWS Glacier-S3 tango was done. The benefits just did not exist unless the client figured out a way to actively make up revenue out the data stored in the cloud. That is the crux of the matter. For every single cloud resource, companies need to milk it to stay profitable. Every corporation is under pressure to ensure that there is trickle down of expenses all the way to their customer for each cloud dollar spent.

The business model must consider resource expenses, monetize people skills, factor in manpower reduction, amortize support costs, value response times, tokenize agility, weight down latencies and in general flatten out the spending curve over a span of 5 years to arrive at a number that can be swallowed whole!

The monkey wrench is of course the fact that cloud architects over compensate and achieve a gross utilization of 20-30% just to handle peak load; unless of course you have built a truly unit scalable distributed cloud application to start with. A closer look at cloud ROI/TCO via a good read is here.

Vendor lock ins are unavoidable

Your favorite vendor has realized that with self service portals doing their work, it is time to ramp up the fight to stay relevant. Cloud journeys have become hotter than Delhi in summer. Cloud development platforms which promise to airlift your on-prem app to the cloud are the flavor of the decade. Core competence seems to be a buzz word that allows vendors to force enterprises to give away large chunks of their engineering to development partners. Multi-cloud strategy looms like the loch ness monster ready to kill your business before it gets off the ground (or so we are told).

Cloud providers have realized this game only too well. They educate and evangelize services to startups, cloud vendors and consultants well in advance of the actual release date. This gives the entire ecosystem sufficient lift-off to put together new offerings to their respective customers. Complex cloud IaaS/PaaS offerings are rolled out giving vendors the sales nous to create more spending for their customers. SaaS is quite troublesome with no inbuilt data migration strategies.

Serverless computing is the biggest lock in that ever exists because it is deeply tied in to a universe of APIs/services/programming defined by the cloud provider.

At the end of the day, the enterprise customer is inextricably linked to cloud APIs, cloud provider services, orchestration constructs, vendor plugins, workload management platforms, business resiliency interfaces and so much more. Containers being the bright spot in this touchy-feely landscape, myriad of dependencies exist ranging from cost calculators à business objects à business processes

Security is being outsourced

Infrastructure security is now a matter cloud provider services like private cloud, threat detection, security policies, URL filters, backhaul sniffers, VPN gateways, etc. As if this were not enough, you have AI/ML driven bots that are able to spread their wings and recon every inch of your cloud footprint. Proactive forensics (like Volexity) infuse you with a Sherlock Holmesian touch of catching criminals who are prowling internecine digital corridors. SaaS services are providing custom security measures for data/credentials/access.

Just like everything else, the enterprise is slowly but surely abdicating security to 3rd parties. It is clear that there still exist many issues that enterprises need to learn about and deal with. But as home-grown skills decline, external cloud experts arise and cloud providers automate monitoring/response; where do we go from here? The multi cloud challenge will make securing data/applications much harder.

Managed security is now the key. Experts argue that with cloud coming of age, how is digital security any different from facility lockdowns? MSSPs (like Verint) are hence born to turbo charge this revolution. The sands of security are shifting and running downhill at incredible speeds. Knowledge gravity is not what it once was with enterprises. The outsourcing of cybersecurity has ramifications across the spectrum of consumptions (application, data, services, resources, user, transactions, economy).

Serverless computing is raising Cain

Serverless has been around as a concept for ages. Google operationalized it in their App Engine platform with instances. AWS eventually commercialized it via their Lambdas. With a $24 Billion USD market, it is all the rage now20. The point is that despite all the good things it does, there are some rather radical fallouts:

• Programmers code less and learn even lesser
•  Cost overruns are guaranteed unless you have a thorough understanding of loads and exceptions
• Application migration becomes akin to disassembling a Harley cruiser and then magically reconstructing a Ducati Panigale out of the same components
• Cloud vendor lock ins are aggravated
• The loss of portability across premises and multiple cloud vendors is no small joke

I would not recommend building workloads with serverless computing unless you have a specific need. I have built architectures in the past for graphical rendering fabrics which make excellent use of short-term computing instances. In many implementations, I have seen more harm than benefit leading to loss of productivity, frustration with billing, constant reworking of architectural constructs and endless fine tuning.

Fort Knox is now a thoroughfare!

I don’t know whether the enterprise cloud qualifies as a bank but the alliteration is just as real. Consider the following:

1. DevOps personnel orchestrating using ephemeral constructs like instant messaging
2. Workloads available on all possible internet enabled devices for audits, primary traffic, testing, development, etc.
3. Distributed control planes that accommodate all manners of accesses, configurations, tweaks and verifications
4. Data as a service creating multiple versions of data in various repositories
5. Multi cloud resulting in application instances spanning premises and cloud providers
6. Authentication service being provided by 3rd parties
7. Geographically diverse access to certain applications and/or data
8. Edge/fog computing creating islands that are lit up on telco highways
9. Co-locations, co-hosting, co-services, co-operations, co-opting?

Enterprises used to have nightmares even after creating fallout bunkers of their datacentres. God knows what they will be contemplating now. Clearly this is a universe that is not for the timid at heart. It is not just a question of security but also one of economics. We are slowly creating connected infrastructure that is opening up the realm to infinite possibilities.

Spiderman swinging his way across Manhattan would surely wonder whether the world has just turned back on itself! Firesales, here we come!

Data gravity is the biggest lead generator there is on the planet

Let us face it, cloud providers are smart operators. The enterprise datacenter market is more than 450 Exabytes roughly split across 16000 companies worldwide. The cloud on the other hand has 1 Exabyte today. IDC predicts that almost 80 Zettabytes will move into the public cloud by 2025 and that more than 30% will be accessed in real time.

What is staggering is the frightening potential of public cloud providers becoming black holes devouring all data in sight. This kind of vortex cannot be fought with mere innovation. The market is slip sliding towards cloud providers at the speed of light. All the cloud providers have to do to control the world is to merely twiddle their thumbs!

Using containers to recoup enterprise data center investments is a pipe dream in the long term. All innovation will happen in the cloud. AI models can be built at alarming rates by just accessing segmented data that should be superfluously available. With business analytics becoming part of every cloud speck, it is inevitable that the market is now wide open. All ideas will be cloud based. Businesses will go where the action is even if they are not particularly innovative themselves.

The world is in the grip of the cloud. This utopian vision is being pushed relentlessly by all parties as if they are staging a worldwide conspiracy. The economics will be so different in a few years when all companies are in the cloud provider’s vice like grip. We can delay the inevitable but not repeal it. Sensible choices don’t exist anymore but sensible voices certainly abound. Maybe we should listen to them a bit and be careful of our cloud journey.

References:

1. https://www.computerweekly.com/news/252476278/A-decade-in-datacentres-how-the-rise-of-cloud-has-hit-enterprise-hardware-and-software-spending
2. https://venturebeat.com/2013/09/05/amazons-mountain-of-margin-in-cloud-services-over-80-profit/
3. https://mms.businesswire.com/media/20191115005367/en/756955/5/data_center.jpg?download=1
4. https://skyline-gazette.com/2020/03/27/global-cloud-storage-market-size-share-growth-trends-and-forecast-2020-2025/
5. https://www.forbes.com/sites/louiscolumbus/2018/01/07/83-of-enterprise-workloads-will-be-in-the-cloud-by-2020/#2bdb5ea56261
6. https://siliconangle.com/2019/07/10/gartner-enterprise-software-sales-going-strong-2019-markets-shrink/
7. https://www.alliedmarketresearch.com/cloud-enterprise-application-software-market
8. https://hostingtribunal.com/blog/cloud-computing-statistics/#gref
9. https://www.cio.com/article/3336274/why-pay-for-tech-certifications-is-declining.html
10. https://www.whizlabs.com/blog/most-in-demand-jobs-2020/
11. https://resources.flexera.com/web/media/documents/rightscale-2019-state-of-the-cloud-report-from-flexera.pdf?elqTrackId=372b6798c7294392833def6ec8f62c5c&elqaid=4588&elqat=2&_ga=2.93643801.522723585.1561405471-1295611161.1561405471
12. https://www.expertguides.com/articles/patent-protection-and-trends-in-cloud-computing/arlntrew
13. https://www.usatoday.com/story/money/business/2018/01/12/worlds-50-most-innovative-companies/1023095001/
14. https://www.wipo.int/wipo_magazine/en/2019/04/article_0006.html
15. https://www.cloudtp.com/doppler/cloud-economics-getting-bigger-picture/
16. https://diginomica.com/rope-me-tie-me-saas-age-vendor-lock-here
17. https://www.infoworld.com/article/3314979/the-forced-march-to-cloud-computing.html
18. https://www.marketwatch.com/press-release/serverless-architecture-market-projected-to-deliver-greater-revenues-during-the-forecast-period-until-2025-2019-10-25
19. https://www.statista.com/statistics/638593/worldwide-data-center-storage-capacity-cloud-vs-traditional/
20. https://www.networkworld.com/article/3325397/idc-expect-175-zettabytes-of-data-worldwide-by-2025.html