What is Infrastructure-as-a-Service and do I need it? Part 2
In this second part of blog series on IaaS, we explore the market landscape for this technology, as well as the benefits, costs and everything else you might need to know to decide your first move.
Who provides Infrastructure-as-a-Service?
Some of the largest IT companies in the world are IaaS providers, and their enormous scale provides low-cost, on-demand cloud computing resources at near-infinite capacity.
Amazon Web Services (AWS)
This is the same Amazon that sells books, clothes and tech gadgets. AWS has arguably the most mature and complete public cloud IaaS services in the world and has been providing them since 2006. It can also be used as a PaaS. AWS has literally millions of customers from NASA and Netflix, to individual software developers and other IT professionals. Storage costs are based on GB per month while compute services are typically charged by the hour.
Microsoft finally unveiled Azure in 2010 and it is available today both for IaaS and PaaS. Azure instances are widely deployed by companies of all kinds, particularly those with major Microsoft investments. Its pricing is comparable with AWS and, like its competitor, touts several hundred different service types for customers to provision.
Google Compute Engine (GCE)
GCE is the comparative newcomer of the big 3 IaaS providers, launching in 2012. Unlike AWS and Azure, GCE hasn’t had to be built from scratch to serve the demand for cloud computing. Rather it is the IaaS component of Google’s global infrastructure – the Google Cloud Platform – that the tech giant uses to run its elephantine Google Search, Gmail, Google Maps, Google Earth, YouTube and other services. As a result, GCE’s scalability is almost unfathomable, though organisations tend to find it slightly more clunky to use than AWS, Azure and others.
Many other companies provide IaaS, notably IBM, Rackspace and VMware. But whereas the giants of AWS, GCE and Azure run on their own proprietary elastic data and compute architectures, the others leverage OpenStack – the open-source cloud computing software platform. In any case, IaaS providers are best judged on their SLAs and uptime, as well as cost.
What are the benefits of IaaS?
IaaS offers a significantly improved cost model for leveraging powerful IT. It also provides the flexibility that organisations demand.
Since the credit crunch of 2008, capital spending has been under greater scrutiny. Capex investments are still positive, where they are merited. In the world of IT, where newly acquired technology can become obsolete within months of buying it, new models of IT consumption have been enabled by the cloud.
Cloud models like IaaS remove the need to own or operate datacentre infrastructure in those cases where it isn’t strictly warranted for compliance or wider data management purposes.
Clearly some legacy datacentre assets that have already been put onto the balance sheet need to be sweated and, in any case, many on-premise private cloud deployments will endure as an essential part of the hybrid cloud mix.
The lack of in-house infrastructure also translates into a lower skills requirement, and this is important when staff costs are high and skills are increasingly rare.
IaaS models not only keep infrastructure-related opex to a minimum, but also remove the distraction of having to constantly refresh core hardware components to the latest generation of performance. This is the responsibility of the IaaS providers, who are constantly driven to increase efficiency and performance by competitive market forces.
Unending on-demand scalability
It is hard to determine the available scale of public cloud IaaS providers but, in practical terms at least, it may as well be infinite.
Literally any IT requirement for additional storage capacity or compute power can be spun up in a matter of seconds and added to the monthly IT bill transparently and predictably.
The knock-on effect of this is greater business agility, enabling effective response to new opportunities and challenges, faster time to revenue of products and services, etc.
IaaS is entirely location independent, meaning there is no need to be in the same vicinity of your data or the infrastructure it is placed upon. Prior concerns about geographic dispersal of disaster recovery sites and so are largely irrelevant, as are risk planning aspects related to the probability of extreme weather events that might impact infrastructure components. With IaaS, the only thing that matters is the SLA.
How much does IaaS cost?
IaaS pricing models are available from the various providers, but until you apply a specific use case, it can be hard to work out what it will cost you to use. These prices are typically equated to small units such as 1GB or one hour, and work out in fractions of a dollar/pound. Needless to say, IaaS costs can quickly rack up. Nobody ever said convenience was cheap.
The variables that impact on IaaS costs include:
● The amount of system memory (RAM) available to a specific instance.
● Amount of storage and speed of storage access (operations per second).
● Amount of processing power (number of processor cores and what kind).
● Duration of use for each instance (how long processing takes to do what you need).
● The need for extra features such as load balancing.
In a private cloud, on-premise IT model, after all those big upfront investments quickly come to an end, and the ongoing costs of maintaining the environment carry on. These costs stay largely the same no matter what your usage is, unless there is a business case to add more infrastructure.
In the IaaS model, you might pay less but you’ll never stop paying. It’s akin to any rental versus ownership comparison. Renting a car for £50 a day looks cheap when the car is brand new and is worth £30,000, and even cheaper when you only need a bigger £75,000 car for a week or so at £150 a day. But run it and service it for 10 years and the total cost of ownership (TCO) figures end up a lot closer than you think, even when the asset depreciates to almost zero.
What are the downsides of IaaS?
Sadly, IaaS is not all milk and honey. There are some downsides, though these rarely mean that an organisation will reject IaaS outright. Rather, as part of a hybrid private/public cloud approach, there will be some IT workloads that warrant a Iaas approach and others that will not.
Reliance upon a connection
With the IaaS instance somewhere out there in the public cloud, your connection to the internet becomes the lifeline to your business. Any disruption to your Ethernet connectivity is therefore going to prevent data going in and data coming out to those instances – both of which could seriously affect revenues, customer confidence and market reputation. As a result, organisations that have increased their IaaS exposure typically make corresponding increases to the size and reliance of their Ethernet services. This naturally equates to higher costs.
Loss of visibility and control
To many, the giving up of responsibility over the physical IT environment that comes with IaaS will be an intended and anticipated blessing! However, this lack of control and visibility presents some concerns around governance, particularly in respect of sensitive data. There are also concerns about who is authorised to leverage IaaS within the organisation, so that costs are appropriately apportioned to cost centres. Lack of fiscal discipline is a hallmark of organisations where the enthusiasm to adopt public cloud services has overtaken governance; a phenomenon known as ‘shadow IT’. Organisations that formally embrace IaaS rarely do so without fully appreciating the reality of this double-edged sword.
Shadow IT is a potential risk of deploying IaaS, though it does reflect just how easy it is to use and just how hard to use traditional IT approaches have become.
It is common practice to initiate some form of governance policy clearly dictating the criteria upon which certain workloads and data sets can be diligently and efficiently evaluated for the suitability for a IaaS environment versus remaining in the private cloud.
Security and compliance concerns
General data protection legislation (such as GDPR and other, more industry-specific compliance) has a significant bearing upon IaaS adoption and the choice of workloads to be applied. The fundamental principle of knowing where data is situated is a challenge that strikes at the heart of the public cloud’s proposition of leveraging a globally-distributed system. The other is the shared nature of IaaS infrastructure and the fact that organisations using it have little or no say over the security systems use to prevent attacks penetrating environments or spreading between them.
What’s the migration path from my existing IT setup to an IaaS model?
Migration to IaaS starts by working out three important things:
1) What infrastructure you currently use
Completing an audit of your existing infrastructure assets gives you a firm understanding of what you might be retiring in favour of a service-based infrastructure model from the cloud. Consider their key attributes such as processing speed and storage capacity, but also things like space, power draw and licensing status. Where are these assets in their natural lifecycle? What management processes and overheads are associated with them?
2) What your current infrastructure requirement is
Determining what you need from infrastructure will be slightly different to what you already have in place. Perhaps you are constrained, or may even have too much spare capacity. Perhaps your computing requirements are met but there are shortcomings in your networking infrastructure or related business continuity/disaster recovery provisions. In any event, this is your opportunity to scope out what you would ideally have to serve your present needs.
3) What future infrastructure requirements stem from your business strategy
IT is a proven business enabler, which makes IaaS a potential turbo-boost to your business strategy. The sky’s the limit so think big about the infrastructure you might need to satisfy the future aspirations of your organisation. What is the ultimate future state of your infrastructure needed to support the next 3–5 years of growth and digital transformation?
These findings should enable you to map out a migration path away from decommissioned on-premise hardware toward an Infrastructure-as-a-Service model.
However, remember that:
● It’s worth trying to switch your mentality away from rack-by-rack thinking to workload-by-workload thinking. In other words, when staging infrastructure migrations into phases, make each stage orientated around a workload rather than a technology. In IaaS, physical infrastructure components are essentially virtual, so you no longer need the baggage of worrying about servers and cabinets in that sense.
● Most organisations will tend to favour a hybrid cloud model that retains some degree of on-premise hardware. So you probably won’t be migrating everything.
● The pay-as-you-go aspect of IaaS is a great benefit to migrations, enabling you to create an entire IaaS environment at very low cost so that you can test and tune it before making the switch as confidently as possible.