The Complexity of Clouds

by Cal Braunstein

Cloud service offerings are like snowflakes – no two are exactly alike.

 

Everybody loves the concept of clouds. Clouds will simplify IT and make life easy for end users and administrators. Amazon, Microsoft, Salesforce.com and others say they will handle business needs so that companies can focus on what is most important – the business itself.

Cloud computingIt sounds ideal – but the reality is less than perfect. From a distance IT clouds can be fluffy and beautiful like real clouds but once one touches it, one can feel the damp chill of reality.

Don’t get me wrong. I like clouds. But before one moves business- and mission-critical workloads to the cloud, one needs to understand the implications and impacts.

Cloud Shapes and Sizes

There are a myriad variety of clouds. Most small businesses take advantage of software as a service (SaaS) clouds – business applications like CRM, ERP, and payroll – but there are also platform as a service (PaaS) and infrastructure as a service (IaaS) offerings. To add to the confusion a number of firms have created new service types such as data as a service (DaaS) and disaster recovery as a service (DRaaS). For simplicity, let’s just look at SaaS offerings.

Just like software, the SaaS offerings from competing vendors will vary according to functionality, look and feel, and the underlying software stacks, systems and data bases. They may appear to be commodities but they are not as interchangeable as cars. Thus, a decision to use a SaaS provider does not mean you can easily switch providers if unhappy with the outputs or service.

For example, if one were using Salesforce.com for CRM and decided to change providers to NetSuite, it could be a long involved effort to migrate over to the new service provider without losing any of one’s data that is being stored on Salesforce.com’s systems. It is analogous to changing banks and credit cards – something you do not want to do often.

Current Cloud Challenges

Users would like to believe that when buying cloud services, you just pay for the service and all goes smoothly from there.

Not the case. Amazon, HBO Go, Microsoft, Twitter and other cloud service providers have almost all experienced extended outages – some lasting for several hours. Most all providers offer no availability guarantees and if they do, it is in the form of a credit for time lost, not revenues lost.

Some cloud service providers (CSPs) have easy to understand contracts that clearly call out what one can expect of the service levels, downtime, maintenance, and outage credits. Others make it hard to determine what the real commitment is or what one needs to do to minimize potential outages and data loss.

Before one signs a cloud service provider contract, one should understand the following components:

 

  • Security – does the security provided satisfy corporate rules?
  • Compliance – does the CSP confirm to regulatory requirements (such as PII, PCI, HIPAA, or SOX) that the company must adhere to?
  • Performance – throughput (response time) and is it consistent under loads?
  • Availability – most offer at best 99.995% (250 minutes of outage a year, excluding the maintenance downtime window)
  • Resiliency – how quick can it recover and from what point in time (what data is lost?)
  • Accountability – what is the guarantee and what are the penalties?
  • Data protection – how is data protected in transit and at rest? Who has access to the data aside from the user? What happens to it when the contract is terminated?
  • Portability – can data or applications be ported to or shared with another cloud provider? If one terminates the contract, how long before the data can be retrieved and ported to a new provider? Will the CSP continue to provide service until one is live on the new provider site?
  • Support – will the CSP provide users the level of support they need when needed? If there are major issues, is there someone to escalate the problem to?
  •