Cloud economics -- Scalability and cost control for CIOs

By Chin-Tang Chin, Microsoft 23-Nov-2010

Chin-Tang Chin, director of developer & platform evangelism, Microsoft HK
Chin-Tang Chin, director of developer & platform evangelism, Microsoft HK
The whole IT industry is raving about the promise of the cloud, which offers the possibility of everything from "infinitely" scalable processing power and storage, along with tools and services that support dynamic capacity adjustment, to load balancing and failover. 

While some vendors only provide infrastructure outsourcing as Infrastructure-as-a-service (IaaS), the true cloud economy of scale is best realized on Platform-as-a-Service (PaaS) offerings.

Beyond these technical IT benefits, what financial benefits does the cloud offer to CFOs? A case study from AXA Seguros show savings from "a one-time hardware investment of almost US$10,000," while another from Siemens demonstrates they have reduced their "TCO by a factor of 10". Let's take a closer look at how cloud is changing the economics of IT delivery today.

Coping with workload burst

For a lot of industry solutions such as online banking, e-commerce, and media production, it is very difficult to predict the demand and therefore the required IT capacity for these sort of services. Sometimes the workload burst is not predictable at all, like the traffic to CNN.com www.cnn.com after the sudden death of Michael Jackson. 

But more often, there is "on-and-off" pattern to the workload spike that is totally predictable. The following diagrams present two good examples, depicting the workload spikes during the holiday season for e-commerce websites and during the tax season for tax software websites (see Figure 1 below).

Figure 1. Workload spikes of different e-commerce websites during holiday and tax seasons
Workload spikes of different e-commerce websites during holiday and tax seasons
(Source: Alexa)

The tendency today is to overprovision data center capacity in order to support strict service level agreements (SLAs). But this leads to extended periods where data center resources are idle and operations overhead is correspondingly high; in this environment, the cost to serve can be artificially inflated. 

The on-demand scalability of the public cloud provides an advantage in those scenarios by offering the ability to bring additional instances of processing applications online or take them offline as demand varies. The flexible scaling option, together with the consumption-based charging model, is the key to driving down wasted resources in data center investments today.








0 reader's comment