Cloud can bloat COSTS. Pardon me?!

By : 0
The Cloud tap looks a lot different when you start thinking of metering, reserved instances, sprawls, over-provisioning, architectures and use-it-or-you-lose-it attitudes. Have you ever said ‘I will buy the biggest one to handle my workloads’? Then you need to read this.

Pratima H

SAN JOSE, USA: Replacing yesteryear’s IT with a glitzy Cloud option is often akin to slipping in a sleek plastic card in the wallet. Letting go of that wad of cash feels so light, nimble, carefree, easy and in vogue.

Unless there comes a day when you wonder why the envelope is so heavy at all; and when you stare at the long bill in utter disbelief, you struggle to decide whether it’s time for a heart-attack or a catatonic fit. Boy! You never imagined a cloud bill can have so many extras and hidden numbers after all. It was supposed to be the other way round. Then why? Or more precisely – HOW!

Understanding the ‘miscellaneous’ in a cloud scenario seems unimportant at first, but ironically it emerges as the most important cat to bell.

Economics is a huge and strong font of any cloud offering’s brochure. But that rests on the premise that the as-you-need-them approach to chewing resources will help you get rid of heavy black boxes and heavier capital expenditure vaults.

Yet, cloud can turn into a poetic paradox when you discover (ah, and too late, at times) that just because cloud is painted with that one stroke called ‘cheap’, your employees, developers, engineers start looking and consuming it with a different mindset. They may not leave switches and taps on when they leave their homes, but the same prudence or frugality does not apply any more to IT buttons, because hey, we are in a cloud. This reckless leaking sooner or later spills everywhere – in wasted computing cycles, application-development neglectful of costs, creation of servers without corresponding deletions, shadow compute shopping and a lot more.

Tellingly enough, while Infrastructure-as-a-Service (IaaS) and Platform-as-a-Service (PaaS) are galloping at a 36 per cent YoY, the potential for cloud waste is culminating to be enormous. Experts bet anywhere between 25 to 30 per cent of wastage in terms of cloud spend happening.

Is it because of self-service going awry, purchasing of oversized virtual machines (VMs), leaving VMs turned on when they are no longer needed, using expensive storage for infrequently accessed data, spinning up unauthorized resources (Shadow IT) or something else or all of it?

Before we reach out for an aspirin or a tourniquet, let’s reach out to Deirdre Mahon, CMO at Cloud Cruiser and see if some kinks can be ironed out when she helps us to look at reserved instances, unit pricing myopia, provisioning policies, metering, monthly-cost mindsets and much more as she zooms in on the small font – the Miscellaneous.

Is it not an irony that cloud investments, that are supposed to be capex-free and hence cheaper, turn out expensive? What do CIOs miss here?

Like any investment or expenditure, it needs to metered and managed. There is no such thing as a bottomless budget or ‘free lunch’. It is true that cloud is an operational cost but there still needs to be a return on that level of investment. Many cloud adopters embrace it because it provides speed and agility which gives you competitive advantage but over the lifetime of that service or asset, it still must be measured and demonstrate concrete value. The harsh reality is that bill-shock happens sooner than you’d expect and the brakes go on and cloud expenditure slows down until policies and governance are rolled out to provide the checks and balances needed.

How much of wastage is happening due to cloud shifts? Why and where in the IT mix is it crawling in more strongly? A case of Inadvertent CIOs or one of helpless CIOs?

Cloud waste can be as high as 30 per cent of overall spend. It comes in the form of over-buying in advance such as a ‘reserved instance’ or enterprise-level account status which requires a minimum commitment of spend a full year in advance. Other examples of cloud waste are over-provisioning of services such as compute or storage and in some cases, services are left running overnight or during the weekend where they are simply not being used.

Can you elaborate on how VMs, storage, shadow IT and monitoring become drivers for this wastage?

Shadow IT is simply a name for when non-IT groups directly procure cloud services. Because cloud rates are charged by unit price by volume and time – all you need is web access and a credit card. Public cloud providers have designed the services to be easily accessed and depending on budget limits, many users can fly below the IT radar. Because unit prices seem so low, users never think ahead about how costs can climb only to discover it when the end-of-month bill arrives. When disparate groups across the entire enterprise self-serve based on their own specific needs, usage can quickly get out of control and we’re not just talking about spending levels, though that is an important challenge. If an organization has no clear rules or policies describing how to provision and access cloud services, it can very easily spiral out of control.

Does this vary when we compare private clouds, public clouds and hybrid clouds? Are customers responsible? Or the vendors? Are there any specific regions, verticals or scale-brackets where customers lean towards wastage?

Yes, governance, control and better management of cloud use and investment does differ by cloud type. Private cloud by its very definition is often subject to greater rigor especially around who accesses it and when. However, in larger organizations where you have multi-users across disparate global teams, it can become more complex and therefore harder to manage week by week and year by year. You still need stringency about how you collect usage and cost data and have the ability to easily map to the business entity and report the relevant KPI’s. Hybrid cloud usage generally means that an organization uses both public and private services together in the same project or application. ‘True’ hybrid cloud is actually rarer and what you often see is organizations leveraging multi-clouds for different use-cases or purposes. Also known as ‘best execution venue’ coined by 451 Research.

At the end of the day, the responsibility of usage and spend tracking must lie with the user. They should expect some degree of reporting and detail on the bill but it is incumbent on the user or customer to track that according to their own set of business processes, rules and overall investment.

Deirdre Mahon

Deirdre Mahon

How exactly can billings get out of control? Is it all about usage or fine prints of cloud agreements? Any sub-text that users might be skipping when going for those huge public cloud discounts?

Billings really get out-of-control because users are enabled to access services easily and quickly. Development teams are not focused on the cost but more on the services type and performance and so they allow ‘sprawl’ until someone who needs to write the check comes asking deeper questions. It’s in large part due to behavior that needs to be better managed but also because it’s a monthly cost as opposed to a big invoice from your former hardware provider which definitely posed lots of questions, extremely lengthy sales cycles with numerous stakeholders at the table including legal and purchasing teams.

Is over/underutilization and provisioning a key facet in the context of wastage? Also, talking of shadow IT, how much of this happens due to poor planning by engineers or perceptions around cloud being cheap or oversight around non-IT users swiping their cards for seemingly cheaper/easy-to-grasp apps?

Absolutely. Utilisation, as I iterate at several points, is key. On shadow IT, yes; much of the waste occurs from the disparate teams of users who have fast, quick access and fly below the radar because of lower level monthly spend allowances.

Bad architecture, hardware mismanagement, legacy baggage, sprawls: what else can be a wastage monster and how?

In addition to the aforementioned culprits of cloud waste, is leaving services running when they are simply not needed. It’s not like you over-provision too much capacity on say storage or you over-commit to consuming networking or security services in say the next X months, it’s really more about being policy-driven and governed as opposed to leaving it the ‘wild-west’ user mentality which enables them to make the decisions. In terms of hardware mis-management – that is really the job of the cloud services provider.

Some waste could be attributed to bad architecture where you are not fully optimized and sized based on what you really need. However, that will become apparent when your usage reports show under-utilized environments or left-over capacity which is generally a ‘use-it-or-lose-it’ for advance purchases such as reserved instance or enterprise account status.

What is the extent of a vendor’s accountability, licensing clarity or tools here?

If yo mean ‘vendor’ as the cloud provider; on that understanding, vendors are accountable for providing accurate and timely usage and cost data so that customers have a chance to monitor how much they’re spending and on what. Some vendors do better than others at providing complete data quickly. Many are at the end of the month showing aggregate or summary views and reports. Vendors’ many licensing options can either help a customer save money, such as discounts for long-term commitments, or leave their users confused as to which service is the best fit.

Confusion often leads to ‘I’ll buy the biggest one to be sure it’ll handle my workload,’ thus bloating cost. The extent of cost management and analytics tools varies by cloud service provider. AWS offers a much richer set of tools than does Azure, for example, with the ability to slice cost in different ways. Generally, these tools or reports are accessed by the account holder and not easily accessed by consumers or other functional stakeholders. This information is not being shared and so users are not being held accountable. This in turn causes mis-use, bad behavior and cloud bloat.

Anything that is worth attention here from examples like the Open Compute reference, G Cloud in UK and micro-architectures (Dockers etc.)?

No. Open Compute: they’re about saving the environment, not saving cost. G Cloud seems to be a provisioning portal for UK governmental agencies to buy cloud services at discount rates negotiated by the government. Great if you’re one of those agencies, but it’s irrelevant to the cloud-buying public. And even if you’re benefiting from those rates, you can still have ‘cloud bloat’ by not using your services wisely. Micro service architectures are technology choices that are appropriate for certain workloads, but not an approach that can be generically recommended to save money.

Tell us something about your offering?

Cloud Cruiser’s application is a smart-meter for hybrid cloud that gives detailed insights into usage and spend, enabling the business to deliver the right services at the right time. By mapping metered usage and cost data with organizational information, customers get instant analytics to optimize cloud investments for business value.

Before we close, what’s the big-picture factor that we can consider when we think of Cloud Wastage?

The reality is that different functions have different levels of responsibility when it comes to leveraging services. End-users care about performance and reliability whereas IT teams worry about security, reliability and types of cloud services. Finance and operational leaders worry about the level of investment and of course value and pay-back or return. Unless all functions work collaboratively and view the same reports on usage and spend, it will continue to pose challenges and unfortunately slows down productivity and innovation.

No Comments so fars

Jump into a conversation

No Comments Yet!

You can be the one to start a conversation.