Excerpt taken from the Akamai whitepaper “Taming the Cloud: Understanding the Risks and Returns of Enterprise Cloud Computing”
As one of the hottest buzzwords in IT today, and often misused, the term “cloud computing” has been the subject of much hype and much confusion. But the potential is real: with its promise of low-opex, zero-capex, on-demand infrastructure, cloud computing offers highly appealing economic and business benefits, such as accelerated innovation and time-to-market, that have given it traction among small and startup businesses, as it gives them low-cost, easy access to true enterprise-grade technology — that could otherwise cost millions to build. For these and other reasons, cloud computing has also drawn the cautious but serious interest of larger enterprises.
Merrill Lynch “conservatively” predicts a $160 billion cloud computing market by 2011, including $95 billion in enterprise business applications.This magnitude makes one thing clear: cloud computing is too important a trend to ignore. Forrester Research agrees, “Cloud computing has all the earmarks of being a potential disruptive innovation that all infrastructure and operations professionals should heed.” Yet while cloud computing holds tremendous potential, regardless of what cloud computing taxonomy one adopts, the cloud is the Internet – and this weak link introduces a number of challenges - particularly within the enterprise market.
Cloud Optimization Services are required to address the enterprise market, and these services must go well beyond Content Delivery Network (CDN) cache-based technologies to remove the cloud-based barriers and accelerate enterprise adoption and realization of cloud computing’s benefits.
How Will the Enterprise Use Cloud Computing?
In addition to using these tiered cloud computing offerings, larger enterprises with existing, mature online channels will leverage a mix of public cloud, private cloud, and origin data center services. The ability to migrate and run components of Web applications across various cloud platforms — based on the business requirements of the application — will be a fundamental tenet of how enterprises will migrate to the cloud. A single site may use IaaS services for storage overflow, PaaS services for custom application modules and best of breed SaaS applications, along with on-premises origin systems. Some enterprises will even establish private clouds, creating a pool of infrastructure resources, typically deployed within their own firewall, that can be dynamically shared among different applications and functions within the enterprise.
Consider that the Internet is the common link between all these cloud computing modules, introducing its specific issues around performance, reliability, scalability, and security…thus, real world cloud computing implementations will include challenges presented by multiple cloud offering integrations as well as challenges inherent to the Internet cloud itself.
Enterprise-ready Cloud Computing Requirements
While cloud computing has gained significant traction among startups and small businesses, Enterprises require the following for cloud computing to deliver on its promise of creating a far more efficient, flexible, and cost-effective infrastructure for their IT needs:
Performance: As enterprises think about shifting from a LAN-sized, on-premises solution to a cloud-based offering, application performance becomes a key consideration. The performance of any centrally-hosted Web application — including cloud computing applications — is inextricably tied to the performance of the Internet as a whole — including its thousands of disparate networks and the tens of thousands of connection points between them. Latency, network outages, peering point congestion, and routing disruptions are among some of the problems intrinsic to the Internet that prevent make it difficult to rely on for business-critical transactions.
Reliability: The numerous recent, high-profile outages at many of the major cloud computing providers highlight the need to provide the high availability solutions enterprises demand, as even small amounts of down time can cost their companies millions in lost revenue and productivity. In addition, wide-scale network problems caused by trans-oceanic cable cuts, power outages, and natural disasters, can severely disrupt communications across large regions of the globe.
Security: Companies worry about loss of control and security when moving applications outside their firewall onto virtual infrastructure — where physical machines and locations are unknown. The Internet introduces new security issues including distributed denial-of-service (DDoS) attacks, DNS attacks, and even application-specific risks such as cross-site scripting attacks and SQL injections. Regulatory and legal compliance requirements present further challenges.
Visibility and Control: Cloud offerings need to provide enterprise-grade support, including robust logging, reporting, and monitoring tools that provide visibility into the infrastructure. Moreover, the Internet, with its many moving parts, presents a complex system to troubleshoot when things go wrong.
Ease of Integration: As most clouds are proprietary, they often require new skill sets as well as re-architecting or re-writing existing applications in order to take advantage of cloud benefits. Enterprises want solutions that allow them to leverage their heavy investment in their legacy applications. This challenge is compounded by the modular, multiple-cloud application solution strategies needed by large enterprises.
SLAs: Service-level agreements (SLAs) are rare among cloud computing providers. And while larger providers offer 99.9% uptime SLAs, this simply isn’t enough good enough for business-critical applications. In addition, these SLAs usually refer to the uptime of the cloud service provider’s own infrastructure, rather than the more relevant measure of availability to end users.
There are aspects of each of these cloud computing requirements that can only be addressed by dealing with Internet issues. To illustrate, while some cloud computing vendors (PaaS providers in particular) talk about providing scale and reliability for their offerings, they are typically talking in reference to the “origin” or first-mile infrastructure that they provide, not to the whole cloud. They may provide automated server failover or a virtual database cluster with automated replication, for example. However, these services are useless against the bottlenecks in the cloud that can adversely affect the end user experience.
This underscores the critical need for underlying Cloud Optimization Services that can tame the cloud — services that will enable cloud computing to reach its true potential.