Before embarking on a cloud journey make sure these basic principles are clear 

Infrastructure as a Service (IaaS) adoption is spreading at a rapid pace. According to Gartner’s findings, the IaaS market has been growing by more than 40 percent per annum since 2011, with revenues expected to reach $34.6 billion in 2017. Looking ahead, the market is again poised to more than double in size and reach $71.6 billion in revenues by 2020. Gartner estimates that by then, IaaS providers will deploy the overwhelming majority of all virtual machines (VMs) as more organisations apply a “cloud-first” strategy.

IaaS offers a shared and highly scalable resource pool that can be adjusted on demand. With full control of the allocated computing power and storage, the operating system, middleware and applications, users can commission and decommission IT resources at a blink of an eye, following a pay-per-use model. The user is responsible for sizing, configuring, deploying, and maintaining operating systems and applications, taking care of backups and so on, which is both an advantage and disadvantage at the same time. While workloads can be set up in whichever fashion needed to suit the individual needs, it does require know-how, time and effort to determine the optimal setup and continuously self-manage the environment. The design can literally change hour by hour, day by day, week by week, month by month as needed, thereby allowing users to swiftly accommodate changing requirements without having to make new capital expenditures, and without losing any time until the new equipment has physically arrived.

In his 2015 whitepaper, John Hales provides a comprehensive overview about things to consider before opting for IaaS. The following list serves as an extract and is by no means exhaustive. Instead, it aims to provide a basic set of questions that need clarification before starting a cloud deployment in order to increase your chances of success and mitigate risks.

[easy-tweet tweet=”IaaS offers a shared and highly scalable resource pool that can be adjusted on demand” hashtags=”IaaS,Storage”]

Networking 

The network layer builds the foundation for all cloud activities. Without it, there is no way of accessing any of the infrastructure deployed in the cloud. Hence, before choosing an IaaS provider, a couple of items should be clear:

  •  What type of access will be required and for what purpose?
  •  How much bandwidth do you need for the planned workloads?
  •  How will the network from your on-premises data centre and other existing cloud providers integrate with the new IaaS provider?
  •  In an effort to avoid duplicative IP address ranges, how does your IP scheme integrate with one of the new IaaS providers?
  •  What speed is possible between VMs and storage?
  •  What is the IaaS provider’s bandwidth between the various DC locations?
  •  How much bandwidth is there within a rack (i.e. to the top of rack switches)?

Compute 

Another critical component is the compute layer comprising the CPU and memory. While there is a whole array of aspects to consider, the following provides a starting point:

  •  Are bare metal/dedicated physical servers available, if needed, or just VMs?
  •  If VMs are available, are they multi-tenant or single tenant?
  •  What are the CPU and memory options?
  •  Are the CPU cores and memory that your VM uses dedicated to you or shared across multiple VMs?
  •  How many CPUs can be placed on a server? Are multicore CPUs available, or are they all single core? How many sockets are available? (Note: These questions have licensing implications!)
  •  What is the memory speed and type? What are the options to choose from?
  •  Can you select the type and speed of storage attached?

Storage 

Depending on the type of workloads you wish to run in the cloud, there are a number of questions on the table. Among them are the following:

  •  What kind of storage classes are available?
  •  What kind of shared storage is available (iSCSI, NAS, SAN, AoE, FC, FCoE, CIFS)?
  •  How does the IaaS provider ensure storage performance?
  •  What is the fail-over concept (active-active; active-passive; failover cluster)?
  •  How is storage billed (space consumed, IOPS allocated, IOPS used)?
  •  What options are available to minimise the cost of static or slowly changing data such as archives or backups?
  •  Keeping in mind the exponential data growth of the digital universe and the fact that data unfolds gravity, what are the available price tiers as you increase volumes and how competitive are they?

Security 

One of the biggest and most cited obstacles when it comes to transitioning to the cloud is security. When considering an IaaS provider, here are some examples of the items that need clarification:

  •  In which location/country is the data ultimately stored, and which laws (privacy and otherwise) are applicable?
  •  Is the DC location in accordance with data privacy or compliance regulations or corporate policies that might be applicable in your case?
  •  How is the DC classified (tier-1 to tier-4), and is this sufficient for your specific requirements?
  •  What certifications does the provider hold?
  •  How does the multi-tenancy concept look? In other words, how is the environment logically or physically separated and secured from all other customers hosted by the IaaS provider?
  •  What kind of secure workloads can be hosted? If you ever wanted a dedicated environment or even caging for highly sensitive workloads, would the provider offer these additional services? Will the provider help you pass audits of those workloads? If so, how?
  •  What access do technicians have to the servers, VMs, and storage used by you? What kind of audit trail is available upon request?
  •  If a security incident were to occur, how would you be informed?

Data Recovery 

Equally important to security is the availability of the workloads deployed. Hence, a long list of items to be clarified with the IaaS provider should include the following questions:

  •  What are the options for local failover (if a VM fails)? Is failover automatic or manual?
  •  What if a site-level failure occurs? How does the IaaS provider ensure high availability across sites?
  •  Are there costs associated for replicating between data centres? If so, are some locations free, cheaper or more expensive than others?
  •  Are there application design or deployment requirements to make DR possible?
  •  What is the availability committed in the IaaS provider’s SLA, and how does this match your requirements?
  •  What is the IaaS provider’s average mean-time-to-repair (MTTR) and how does this compare relative to other IaaS providers or industry standards?
  •  For business critical workloads: What is an appropriate penalty in case of a major service interruption, and is the IaaS provider willing to agree to these terms?