Cloud Computing Challenges and Risks

This post complements the post Cloud Computing: Hype, Vision or Reality?

The previous post describes the presentation by Pini Cohen STKI Vice President in a in the 23^rd meeting of the SOA Forum of the Israeli Association of Information Processing.

This post describes the interesting discussion during the meeting, which was focused on Cloud Computing challenges and risks.

Security

Often cites as the major risk of Cloud Computing and verified as the leading cause for avoiding Cloud Computing in surveys. One of the participants (a Security expert) argued that Security is partially real issue and partially psychological issue. The image of less controlled environment (Cloud is less controlled by a customer) is an image of less secured environment; therefore we should expect resistance to Cloud Computing.

My take

The issue is not only Security. It could be also an issue of Privacy.

For example, usage of Infrastructure as a Service (IaaS) for backups by a home consumer may expose his personal data.

Availability

Planned Cloud unavailability schedule is not controlled by customers. It could be in a customer peak hours due to the global distribution of the Cloud customers.

My take

• I do not think that planned unavailability is a real issue. Clouds Providers will minimize planned unavailability time and restrict it to parts of the cloud, similarly to the high availability of current Data Centers.

• The realistic concern is unplanned service outage. Most, if not all Cloud Computing providers experienced unplanned outages. For example Saleroce.com outage on January 6^th 2009.

• Another problem is a difficulty to diagnose the problem's root due to multitude of technological components and layers (including communication via the Web and Virtualization). The cited above SaleForce.com unavailability incidence reason was not identified during service outage time.

Tasks for Clouds and tasks for Internal Clouds

Services types suitable for inclusion in web based Clouds:

• Autonomous (independent on other Services) e.g. employee evaluation

• less critical services

• Development and testing with scrambled test data.

• Training

Internal Clouds would probably execute production workloads e.g. Legacy applications (including Business differentiating functionality).

SaaS Adequacy: CRM & ERP only?

The issue is Cloud adequacy for hosting Legacy and proprietary applications. These applications maintenance is a lot more difficult than maintenance of pre-built applications such as CRM and ERP.

My take

• Currently even customization of pre-built applications is an issue. CRM is the most prevailing SaaS implementation. Is this the reason why even ERP solutions are relatively new in the Cloud?

• It is part of a broader concern: building a Virtual Enterprise composed of a Data Center and multiple clouds. Functional Integration, Business Intelligence and EII are other issues in the virtual enterprise context.

• As far as SOA is concerned, location of Services or parts of Services is more significant than within an enterprise.

Infrastructure vendors: Is Cloud Computing a threat or an opportunity?

Infrastructure vendors' revenues could decrease significantly in Cloud Computing model Due to the efficiency of that model (efficiency is less servers,  less licenses and less licenses fees).

Vendors like IBM, HP and SUN have to resolve a conflict between adopting the Cloud model as Cloud suppliers and cannibalize their revenues or maximize revenues of current Data Centers based sales revenues.

As a compromise between these approaches they offer to large customers Internal Clouds model. Some of the participants argued that internal Clouds as Utility Computing and not Cloud Computing.

They could also innovate offering new Cloud Based services and products.

My take:

• Internal Clouds models could lower costs significantly. They address fears and doubts and resistance to the new Cloud Computing model. They are not adequate to SMBs which will adopt Cloud Computing earlier than large enterprises.

• Large enterprises will have to follow consumers (already adopted the Cloud model by using services like gmail) and SMBs. Therefore for the Long Term Infrastructure vendors should support the Cloud Computing model.

• The Internal Cloud Model will coexist with Web based clouds.

• Licenses revenues will decrease not only because of efficiency. The other reason is size. Cloud Computing suppliers size will drive price pair unit (hardware & software) down. The lower hardware and software licenses price to Outsourcing giants is an indicator of future prices trends of sales to Cloud Computing giants.

• Reduction of prices will not be limited to Infrastructure products. It is also applicable to applications (ERP, CRM, SCM etc.).

Which vendors will dominant Cloud Computing for the Long Term?

Current Infrastructure vendors (IBM, HP, SUN, Microsoft etc.) will probably dominate this market but expect innovative niche solutions by new market players, for example in Integration between Clouds and Data Centers.

My take:

• Web 2.0 vendors like Amazon and Google will be also dominant players. Consumers will probably prefer these current Cloud Computing leaders over the traditional infrastructure software and hardware vendors. Probably they will compete with infrastructure vendors in the SMB and Enterprise markets.

• Cloud Computing is a major paradigm shift. Traditional infrastructure vendors should adapt their Business and Organizational models to the new IT world, similarly to the change from Hardware centric model to Software centric model more than 10 years ago. My prediction is that not all the infrastructure vendors will succeed (like DEC which failed the transition from Hardware centric model to Software centric model.

Hints about my predictions which vendors succeed are available in my vendors-survival posts on: SUN, HP, Microsoft .