When it comes to the cloud sometimes navigating the many terms, jargon and advertisements can be confusing. This cheat-sheet summarizes some of the key concerns when it comes to cloud.
What is the cloud
Cloud is commonly used to refer to the outsourcing of software and IT as a service or utility.
Within the context of infrastructure this can mean outsourcing the management network, storage, physical servers or to even applications to a 3rd party.
When does it make sense to use the cloud?
Most business decisions are informed and take an economic view. Any given solution to an IT problem can be seen in the context of “required initial investment” (CAPEX — capital expenditure) and “operational costs” (OPEX operational expenditure).
The first approach of an on-premise solution is initially high CAPEX followed by low OPEX. Cloud based solutions, relative to on-premise solutions offer zero CAPEX with a low initial OPEX. However as demand scales so do the costs in a linear fashion.
Given static goals such as fixed demand over time, in most cases, the high CAPEX approach always yields a better investments.
Given variable demand, in most cases, a high OPEX approach allows for Just-In-Time infrastructure costs to be tightly aligned with usage demand.
SLAs such as availability and mean-time to recovery also influence decision making. The costs of DDoS protection for example range wildly between CAPEX and OPEX based solutions.
Finally economies of scale mean that cloud providers can always offer lower initial OPEX based solutions. However once scale reaches a certain point CAPEX based solutions begin to reduce costs for a company.
When does it NOT make sense to use cloud?
Given static demand over time, in most cases, not using the cloud always yields a better investment. Given a large amount of storage or compute resources the OPEX costs of the cloud may out grow in comparison an initial CAPEX approach. Some companies are simply too big for the cloud.
However many companies still choose to sacrifice high OPEX for increased business agility. For example Lyft.com (alternative to Uber) spends $80 million a year on OPEX cloud costs. Netflix spends $40 million a month.
Projects like the CERN particle collider focus on high CAPEX solutions as the priority is on low operating costs.
Given a business problem, a company’s investments and cash-flow, there is a so called “sweet spot” where the right combination of CAPEX and OPEX makes sense.
Most successful start-ups focus on an OPEX approach until demand shows a transition to a CAPEX approach makes more sense in the long term.
What is private cloud?
Private cloud is the ability to deliver a suite of unified cloud technologies and run them directly in your data center. Private cloud is typically a combination of IaaS (Infrastructure as a Service) and PaaS (Platform as a Service) capabilities.
Private cloud service providers typically don’t charge for upfront hardware costs. They offer per usage billing as they profit from your long term usage costs. Private cloud solutions typically encompass a unified set of solutions around billing, monitoring, Identity and Access Management, provide a service catalog of on-demand managed services and via self service portals increase a company’s agility through organizational autonomy.
This is different from traditional on-premise visualization investments. High inter-department coupling, due to unconnected products, reduces business speed and agility. Private cloud is an evolution from traditional silo-ed virtualization solutions.
What does “re-platform” mean?
There becomes a point in time when the technology investments made yesterday can’t meet the business goals of tomorrow. Globally, companies re-platform on average every 5 years however this is occurring more often as the market becomes more competitive and disruptive.
The strategy towards re-platforming differs whether a company is CAPEX or OPEX orientated. A company with a high CAPEX DNA might re-platform in the form of raising investment. A company with a high OPEX DNA typically absorbs re-platform costs within OPEX costs in exchange for agility.
What are the different solutions available to a company?
When it comes to the infrastructure there are a lot of investment options available. The solutions below highlight the spectrum of business postures a company can take with respect to infrastructure and IT. The extreme left reflects high CAPEX low OPEX options and the extreme right reflects high OPEX low CAPEX options.
- Traditional on-premise IT; full control, dedicated data-center. High CAPEX low OPEX at scale.
- Co-location, a more economical and high-medium CAPEX take on opening a data center with the support of a partner. However all other parts of the solution are remotely managed.
- Hosting, typically a conservative evolution from a company looking to outsource networking, storage and server provisioning expertise. Expertise in managing visualization appliances remain in-house. OS patching and software installation expertise are also maintained in-house. Typically a multi-year contract is negotiated with medium CAPEX and medium OPEX commitments. A great choice for non-variable demand.
- Infrastructure as a Service (IaaS) also known as private/public cloud. Low CAPEX and pay-per-usage OPEX. IaaS customers are typically coming from previous hosting investments and are looking to scale and automate organizational processes around firewall, storage and compute allocations.
- Platform as a Service (Paas). Most IaaS solutions include PaaS capabilities. Therefore PaaS and IaaS investments are interchangeable. PaaS offers the introduction of a managed application catalog allowing companies looking to outsource the expertise of managing database applications, communications or intranet applications to the solution provider.
- Software as a Service (Saas) outsources the entire application life-cycle of a product to a service provider. Examples include email, telephony, CRM, financial and document storage.
What is “server-less”?
Server-less is the ability for a service providers to bill per HTTP request. Typically in IaaS or Paas setup, a minimum set of resources are provisioned to meet capacity. As such a company might be billed per hour per CPU even when there is no application traffic.
With server-less, an application can be packaged in such a way that when HTTP requests comes in, CPU and RAM resources are dynamically provisioned. Take the example of a weekend with no resource usage:
- Data center — “keeping the lights on” costs incurred
- IaaS / PaaS — minimum CPU and RAM costs incurred to run the application in a low usage mode (e.g. database and monitoring servers)
- Server-less — zero costs incurred. Only as HTTP requests arrive, resources are dynamically provisioned
An article by : andrew webber