Session abstract: “The relevance of cloud computing to enterprise IT is not well understood. Like the Internet, cloud computing did not come out of blue. Dr. Zhou’s keynote will discuss the historical evolution of cloud computing including the adoption of grid computing to share IT resources in large enterprises. This presentation will focus on internal/private clouds suitable for most business applications utilized in an enterprise environment. Dr. Zhou will also describe the evolutionary steps required to adopt cloud computing and the potential hurdles to be overcome. Using production cases, he will discuss how cloud computing is not only saving costs but also driving competitive advantage.”
Songian Zhou is a co-founder of Platform Computing. This session focuses on establishing a private, infrastructure-as-a-service cloud. What follows are notes I took during the session, and some screen shots of Dr. Zhou’s slides.
Dr. Zhou begins with cloud computing benefits, expresses his view that cloud computing is the third generation of IT, and highlights three pathways to cloud computing.
Why should we bother with cloud? Faster, cheaper and better: on-demand delivery of processing power, improved performance of business applications, increase utilization through resource sharing, reduce CapEx & FedEx, improved quality of service for end users, self-service resource procurement.
Cloud computing is third generation of IT. Generation 1 – Mainframe / Centralized computing (“old-time cloud”); Generation 2 – client/server (networked computing), cloud computing – distributed computing (internet wide environment, services model).
Next, Dr. Zhou describes when an organization might use public vs. private clouds:
Public Cloud
- non-mission critical SLAs
- in-house IT has limited scale, scope or expertise
Private Cloud by Corporate IT
- maximize value of underutilized resources (compute, not people)
-mission critical SLAs
- high security and compliance requirements
- enterprise specific services
He points out a frequent adoption scenario, in their experience, is customers starting with a Private cloud and then (in controlled manner, via gateway) burst additional workload to Public cloud.
In setup for his discussion on IaaS and Private Clouds, Dr. Zhou points out 3 Level of Cloud Services:
SaaS – application for end users on demand
PaaS – APIs for developing new Cloud applications
IaaS – server, network, resources
Dr. Zhou shows 3 pictorial views of Private IaaS clouds: stack model view, cloud services view and usage view.
During this portion, Dr. Zhou emphasizes having a services, rather than underlying resource, point of view. IT is offering a service to its constituents. The end view of that service is an application. In the usage view slide, the application manager is what some would refer to as a service owner or product manager.
From here, Dr. Zhou speaks of the Misconceptions of Private IaaS Cloud:
- All resources in company must be managed in the cloud; false, pick and choose applications
- I need to replace existing servers; false, can reuse existing servers
- I need to change my entire IT organization; false, can evolve over time, leverage existing infrastructure teams for both
- I need to change all my IT processes; false; cloud operates under existing regulation and compliance: builds, directory & security
Although Dr. Zhou says ‘false’ for each of the above, each has the “over time caveat”, over time, everything will change, so each “misconception” becomes a truism (obviously, my observation, not his direct statement)
Next, which applications are suitable for the cloud. There are three primary factors:
- location; sensitivity to where application runs, more sensitive (dependent) less of a fit
- workload: predictability and continuity of application load; is there existing dormant capacity to be utilized? is the workload bursty? both are good candidates; continuous workload (fully utilized already) is not a good candidate
- service level – severity and priority of service level agreements; more critical, less likely candidate
He shows examples that are not viable: customer portal, email gateway, clearing house, bank messaging and market data. Interesting on the customer portal. Sounds too general of a category to me. I can see the others.
Next, common enterprise requirements for a private cloud:
Finally, Dr. Zhou closes with some customer examples of IaaS private cloud computing, including Citi’s use of Platform products to create an enterprise utility – shared resource grid; JP Morgan Chase creating a corporate compute backbone, an unnamed bank doing global data collection for a data warehouse environment, and a Singapore government grid/cloud initiative (alatum).
Comments