Implementing Cisco Data Center 3 · Another benefit is the overall ‘quality’ of the configuration resulting from autiomation (reducing the number of incidents post configuration),
IT’s customers are predominantly the workforce of Cisco, it’s Partners, our Diversified Businesses (WebEx, LinkSys, Ironport etc.) and their contingent workforce. Like other businesses, our customers can broken up into market verticals (i.e. Sales, Marketing, Manufacturing etc.) Using XaaS terms becoming somewhat standardized in the Cloud vocabulary, the various layers in the IT stack can be converted into SaaS, PaaS and IaaS service offerings. IT must understand the relationships of this entire stack for two main reasons. Operationally, we need to understand if we change some lower in the stack, what impact will it have on a Customer facing service. Financially, we must understand the cost contribution of each layer in the stack to the TCO of a customer facing service. We need a common SMF to manage all layers of the stack in order to obtain that comprehensive view. Service Level Management Service Costing Capacity Planning Problem and Incident Management Asset Management Configuration and Change Management Provisioning Entitlement Service Catalog Operational Model Evolution Service management framework
Virtualization has created a market transition. Servers are becoming fluid objects in the network. Cisco IT is developing an elastic architecture alternative to traditional add-on approaches for data centers.
…
Presenter
Presentation Notes
Data Center Evolution Path to business alignment: The biggest IT issue today is a lack of flexibility, which is exactly what is required to keep pace with rapidly changing business conditions. Today’s Data Center optimization initiatives are key catalysts to bringing: Consolidation for standardization Virtualization for asset optimization Automation for improved service delivery. Beyond automation, you begin to enter the next level of IT refinement with robust utility pay-as-you-go services offered within: Enterprise intranet-class cloud utility services and Inter-cloud go-to-market value deliver chain services for development and delivery of customer consumables. Virtualization is changing the DC architecture Unified computing is an architecture focused at the driving cohesion and simplification to our customers on this journey.
NO-cannot share with customers 3 scenarios: A on the left, B in the middle, and C on the Right Configuration (actual): non-Cisco 3rd party rack mount server, traditional Cat6k distribution & Cat4k Top-of-Rack acess layer, no virtualization. Physical unit TCO is $3,579/quarter. Configuration (actual): non-Cisco 3rd party rack mount server (as in A), traditional Cat6k distribution & Cat4k Top-of-Rack acess layer (same as A), real-world virtualization adoption for a large enterprise (Cisco IT, RCDN9 facility, 54% of OS instances virtualized as of Dec 1st, 2009). Average virtualization platform is around 2x4 Silver (2 cores, 4 GB memory, 50% resource reservation), at $1,025/quarter. Configuration (realistic target): Cisco unified computing system, Nexus 7000 network switches, Unified Fabric including Nexus 1000v in DMZ, server bay design (2x switch for 10x UCS chassis), Cisco IT Elastic Infrastructure Services (CITEIS) automation effort (automation target Q2FY11), 3-yr cost absorption plan for automation, automation enabled for virtual and bare-metal, virtualization increased to 59% based on (i) expanded memory footprint from UCS, (ii) adoption of virtualization enabled by Nexus1000v in DMZ. Unit costs drop to $2,592 and $742 for bare-metal and average virtual respectively. Common assumptions: Identical discount rate for non-Cisco 3rd party rack mount server and Cisco UCS. Identical CPU technology (Nehalem) and memory technology assumed in non-Cisco 3rd party compute hardware and Cisco UCS hardware. Costing based on real-world Tier-III facility (5.4 MW to the IT floor) actual spending, adjusted for like-for-like discount rate. Compute unit TCO includes: facility, power, equipment, software, automation, implementation and operational cost. Service offering scope includes: compute, core network, distribution network. Scope excludes: storage consumption, WAN, ISP connectivity. Capital cost absorption matching hardware lifecycle time of 3 yrs. Story: Traditional platforms without virtualization have a quarterly unit TCO of >$3,500 / quarter. Real-world execution of virtualization without changing the hardware or the design leads to a 39% reduction in average unit TCO. The adoption of unified computing and automation leads to an additional 32% reduction in average unit TCO. This comes from a number of factors: (i) unified I/O leads to more efficient usage of the overall infrastructure hardware (network/SAN), (ii) automation significantly reduces implementation and operation cost, (iii) expanded memory and Nexus 1000v in the DMZ result in increased virtualization rate Note that the virtualization rate does not stop at 59%. As technology keeps maturing and application vendors adjust their support agreements and license models, the trend will continue. Additional benefits: compute service provisioning agility. Moving to a self-service model substantially reduces the end-to-end delivery time. Cisco IT compute service consumers consider this to be a major benefit. Another benefit is the overall ‘quality’ of the configuration resulting from autiomation (reducing the number of incidents post configuration), and the resiliency for both planned and unplanned events enabled by virtualization. Longer-term cost benefits are expected from the reduction of high-end custom compute platforms (non-x86) onto x86.