[Adopted from my BLOG December 2009]

Lately in the IT community all the hype is on Cloud Computing. We have small start-ups offering several variations of Cloud services as well as some of the established big players (Google, Amazon, IBM, Novell (aimed at cloud service providers),Sun) stepping up their offerings of cloud services.

But what exactly is Cloud Computing? Is it Virtualization? Is it services that we accessed via a web browser over the years, something totally new, or is it all of these,but  just rebranded?

The term Cloud Computing started gaining traction when Google and IBM launched a university initiative to address internet scale computing back in 2007.

These services has been evolving since the 90s and its previous incarnations can be said to be Grid and Utility computing and the Software as a Service offerings we saw around a decade ago.


Cloud Computing Journal - the Web's most widely Web resource on Cloud Computing

In a nutshell we can draw an analogy which can be stated as this: think of a utility service you use, say for example electricity. You get your meter read every few weeks and you receive a bill for energy consumed between readings.

The same underlying premise can be applied to a cloud service, an end user can subscribe for any of the offered cloud services and based on service usage from the provider be billed for consumption of that particular service or series of services for its specified time-frame.

Once can safely state that Cloud Computing as an on-demand, self-service, pay-as you go utility, evolved from a combination of grid computing, virtualization, and automation.

Experts estimate that this industry will grow to a 42 billion dollar industry by 2012, however the implementation and usage of cloud computing models and services is not without issues.

Most business managers will most likely consider the Capex and Opex aspects, especially in this economy. How much money an IT department can save yet still maintain operational efficiencies and security is a primary focus; by implementing one or more cloud computing services, an enterprise can obtain the scale and flexibility it needs and potentially save time as well with the concepts of dynamic provisioning of needed services.

One Cloud Computing claim is to lower costs, increase business agility and help increase the velocity at which applications can be deployed, however a good question to consider is can one expect its implementation to be disruptive and to what length?

In order to engage cloud computing services, business models will have to be adjusted or downright changed, in order to effectively and efficiently managing the utility aspect of computing power used in everyday operations and the manner in which management will be able to utilize resources.

As with any implementation, standards and regulation needs to be formulated and implemented in order to ensure that both vendor and the tenant are in compliance and within governance of an agreed format of policies.

As of now there are no formal standards directed solely toward cloud computing however NIST has proposed a potential framework standard called Standards Acceleration to Jumpstart Adoption of Cloud Computing (SAJAAC). With this, every effort should be made to ensure the confidentiality, availability and integrity of data held within a cloud computing environment going forward.

Definition:
The National Institute for Science and Technology (NIST) defines cloud computing as "a pay-per-use model for enabling available, convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, services) that can be rapidly provisioned and released with minimal management effort or service provider interaction." [2]

Cloud Computing Models: Cloud models can be one of the following three:

• Infrastructure as a Service (IaaS) e.g. Tier 3, Amazon EC2,while the subscriber does not control the cloud infrastructure they do have control over select portions of network e.g. firewalls, operating system, deployed applications and storage.

• Platform as a Service (PaaS) this goes back to the early 70's when it was referred to as Framework as a Service. What is does is simply to provide different combinations of services to a subscriber supporting an application development life-cycle e.g. Google's App Engine which will let a subscriber run web applications on Google's infrastructure or Azure. In essence the subscriber will use programming(.Net Java python) and tools supplied by the service provider with no underlying responsibility for the cloud deployed network, severs, operating system and storage etc.

• Software as a Service (SaaS) e.g. Facebook, Salesforce.com, applications running on a cloud infrastructure that can be accessed via a web browser interface.

Bear in mind that there can be dependencies and a relationship between the models as Infrastructure as a Service can be stated as the foundation of Cloud Computing services, upon which Platform as a Service and then Software as a Service is built upon.[3]

Delivery Methods:
these services can be implemented by the end user in four different manners :

1) Private Cloud aka a corporate cloud refers to proprietary computing architecture providing hosted services to a limited number of people which resides behind a corporate firewall, in other words a single tenant.

By using private clouds enterprises/tenants will receive the same economies of scale and bi-directional scaling as that of the public cloud user.

However being a single enterprise or division within that enterprise will ensure the additional benefit of more control and security for data held within the private cloud, as on-premise data centers can be converted into private clouds by implementing virtualization technologies from companies such as Microsoft, Citrix, VMware, Novell and Sun.

2) Public Cloud is the cloud that is provided for lease external to an entity's physical location e.g. Amazon's EC2.

This deployment facilitates rapid scaling via virtualization technologies (which enables cloud user resources an ability to rapidly start up and shut down,) and can be utilized by multiple tenants however; within this deployment users have no access to dedicated resources.

This results in users giving up a certain amount of control over the process, which in turn can raise security and compliance issues.

3) Hybrid Cloud is a mixture of the public and private. This can be appealing for a company that chose to store non confidential data externally say using Simple Storage Service (S3) whilst keeping private data in-house.

4) Managed Cloud -In this offering the physical infrastructure in operation is owned by the subscriber and can be housed within the physical premises of the subscriber.

However the service provider will control portions of management and security of the service utility.

Some Deployment Concerns
As with the deployment of any IT system there will be challenges and cause for concern. Certain scenarios will have to be anticipated and use cases as well as processes to mitigate these concerns need to be clarified. Some examples of concerns are as follows:

1) Software licenses: software is typically subscribed as those that are proprietary licensed or those that are free and open source licensed. Software licenses govern usage or redistribution of software which are in most cases copyright protected.

Something to consider is how software that you want to deploy into a cloud is licensed.
Is the software you want to deploy licensed on a per server basis or not and how will easy or difficult will deploying your software of choice into a cloud be?

Will proprietary software solutions need to be confined to dedicated hosting environments? Most likely yes, at least for now or until you can get a vendor who can securely provide the software you need on a pay as you go basis.

Because of how software licensing is structured early cloud users have been found to use more open source software.

2) Single point of failure: a mission critical application is deployed via a single vendor; issues at the vendor's site may severely impact the availability of resources for the tenant.

The vendor may claim to have multiple, remote backup locations completely powered However when it comes to ensuring that, in addition, redundant cloud administration and infrastructure software are in place, the vendor may fall short.

3) Portability: a cause for concern is that each vendor may utilize different applications APIs and formats for data. This in turn may limit application and data portability to other environments, as they are likely to be using proprietary APIs thus causing a "Lock In" situation where it will be easy to sign a contract and use a vendor service however transferring out to another vendor could have major issues.

Of course as the cloud computing environment evolves this may soon be remedied.

4) Security: the elephant in the cloud room. The most common fear with using a cloud deployment is a loss of control and security of data.

Granted this is still a system built on hard and software platforms and as such is still susceptible to the traditional security attacks (DOS, DDOS etc.), conversely a point for consideration should be that any security measure will be more cost effective when implemented on a larger scale.

Any good IT manager has voiced concerns over whether employees/administrators at the cloud provider can be trusted to not look at data or even modify it or, whether other customers sharing the cloud can hack data or access it without leaving an audit trail.

From this a tenant can ask about methods the vendor is employing to protect data such as high physical security as well as what types of monitoring, intrusion detection and firewall equipment are in place at their centers.

Even worse is whether competitors could find out sensitive information such as customer orders, pricing and cost information, and negatively impact business. And of course what about privacy concerns and government regulations?

Other issues of concern can be:

• What levels of protection in place to protect one customer from accessing another customer's data or application within a shared cloud space?
• Who will be liable for security breaches and how will the law regarding this in any one jurisdiction ensure compliance?
• How well will a vendor system integrate with a tenant's security systems?

5) Scalability: Every user/potential user of the cloud constantly hear of the substantial savings they will realize by utilizing cloud-based resources.

In order to take full advantage of the scalability of the cloud there should be a means of ensuring that there is some form of dynamic measurement and resource management for applications held within a cloud.
Scalability within the cloud can be had by composing the service from other scalable services as can be seen with Google App Engine.

6) Auditing: With the cloud one has to consider how compliance with ISO standards, Sarbanes-Oxley, HIPAA ,PCI-DSS etc. will impact certain data from being deployed. This more so when considering the attractiveness of data to unauthorized entities and the methods they could use to gain access to that data.

Any IT manager will also tell you that without proper planning the cost of an audit can be higher than expected.

As of this writing, I am not aware of any formulated standards for auditing within the cloud, however I must state that for a business, auditing within the cloud may be an attractive option as, this can be done live with no down time or interruption to business processes.

7) Compliance: There are no standards in place as of yet, but the National Institute of Standards and Technology (NIST )and others are working toward that end.

8) Other Data Access: what happens to my data if the vendor revokes my access or there is a system malfunction? This is a common question that anyone thinking of using the cloud should ask. Remember the data loss for T-mobile customers using the services of Microsoft subsidiary - Danger? "Microsoft said any data that users had on their devices and is no longer there has almost certainly been permanently lost"[4] Here there was no revocation of access but an alleged system glitch.

A personal example occurred a few days ago when I tried to access a Gmail account I kept just for research and online backup.
The system message intimated that I had violated the "Terms of Agreement". What?! The Gmail account was hardly ever used to send email and the Google docs account was used as a second online backup for some of my documents and files. If this was not a secondary backup or not a backup at all, I would have lost access all my uploaded documents and files, with no recourse for resolution but filling out a form and hoping for contact from the support center.

In order to address and mitigate these issues the tenant should ensure that workarounds and backup plans are worked into their Service Level Agreements (SLAs) with the vendor.

Service Level Agreements (SLAs)
A service level agreement a part of a service contract where the level of service is formally defined. In practice, the term is sometimes used to refer to the contracted delivery time (of the service) or performance. [5] Whilst there may not be much flexibility with a vendor in defining an SLA, I am confident that the laws of supply and demand will shift this more toward the tenant in the near future.
Cloud computing vendors are getting into this business to affect their bottom line and shareholder value if publicly traded. At the end of the day their focus will be on making a profit on services offered.

In light of this most tenants may feel as though they are getting into an arrangement where it appears as though vendors create the SLAs for their own protection against litigation, with minimal assurances to a tenant.

That being said, this does not mean that an IT manager cannot make the SLA work as a tool to chose an appropriate service provider. An IT manager's main concern will be the security of data and of course, the traditional interpretation of the CIA triad (Confidentiality, Integrity and Availability) may not be applicable within their cloud service.

To start an IT manager can focus on the following when hammering out their SLA with a vendor:
1) Data Protection: where there is a clear definition as to who will have access to the data and the levels of protection in effect for their data.

Some questions that can be asked are:

• How will data be encrypted?
• How will compliance be addressed?
• What are the levels of access control?
• Will there be sub-contractors or third party providers processing the data?
• Where are backups stored?
• How is the data center secured?
• What happens to the data if service providers are switched?
• What processes are in place to mitigate legal inquiries about a customer's data?
• How often are audits done and what types of auditing tools are in place?
• What happens to my data if there is an investigation taking place on another tenant sharing services and how will you ensure my access to my data in the event of equipment seizure by federal entities?
• How is data deletion handled?

2) Continuity: one has to consider what happens in the event of an outage or another related event that causes data to become unavailable.

Some questions to consider here are

• How will the vendor define a services outage?
• Will there be scheduled vendor downtime for maintenance etc.?
• Will there be an alternative vendor hot site or vendor site prepped to take on load of access in the event of a vendor outage?
• Are there tools in motion which will determine the severity of a vendor outage?
• How will the tenant be compensated in the event of a vendor an outage?
• Define levels of redundancy in place to minimize vendor outages?

3) Costs: on cost to consider are:

• How is the vendor's fee structured and is taxes and external fees accounted for in a vendor quote?
• Will there be or are there current licensing fees above and beyond stated vendor service fees?
• Will there be any hidden or add on costs for vendor support?
• How does the vendor structure their charges? Is it based upon usage, traffic or storage limit
• Does the vendor offer price protection?

Conclusion
It is expected that Cloud Computing will the wave of the future in terms of computing, it is only logical that the cloud's economies of scale and flexibility will impact how technology evolves and how users of technologies implement these technologies.

However in terms of security the massive availability of resources and data within a cloud does present a very attractive target for attackers.
That being said, we can assume that cloud-based defences may be more robust, scalable and cost-effective, in an effort to mitigate security concerns regarding multiple tenants, encryption, trust and compliance.

Part of a cloud service is the API. However when it comes to integration between vendors this may pose a problem for tenants, given the fact that cloud APIs are not yet standardized. This means that each vendor has a specific APIs for managing its services that will lock customers to their vendors due to vendor proprietary technology.

The work around here would be to look for vendors that use standard APIs wherever possible. This is a viable option as standard APIs are already implemented for access to storage as well as deploying and scaling applications.

In terms of auditing and forensics, dedicated, pay-per-use forensic images of virtual machines can be obtained by an auditor without having to take infrastructure offline. This of course results in less down-time for auditing as well as it can provide cost-effective storage for logs without deterring system performance.

All of which will increase the return on investment as well as decrease operational costs normally involved with in house systems processing the same data as in the cloud.

Of course Cloud Computing is still in its infancy and whilst some proposals may look good in theory, only time will tell how we proceed and evolve with this system of computing.

References
[1] Cloud Connect
[2] NIST
[3] Cloud Security Alliance
[4] www.pcworld.com/businesscenter/article/173457/microsoft_loses_sidekick_u...
[5] http://en.wikipedia.org/wiki/Service_level_agreement

Cloud Computing journal
European Network and Information Security Agency.
Sun Microsystems
IBM
Google.com

• Welcome 2010 & Cloud Computing
• A New Generation of Secure Cloud Computing Solutions
• Open Source & Cloud Computing: Is There a Free Software Alternative?
• Standards Acceleration to Jump-Start Adoption of Cloud Computing