Professional Documents
Culture Documents
in Datacenter
Panagiotis Golemis
Consulting System Engineer
Datacenter Technologies
Internet of Everything is Here
Networked Connection of
People, Process, Data, Things
People Process
Delivering the right information
Connecting people in more
to the right person (or
relevant, valuable ways
machine)
at the right time
IoE
Data Things
Physical devices and objects
Leveraging data into more
connected to the internet and each
useful information for decision
other for intelligent decision making
making
Complex IT Processes to Deploy Application Resources
Challenges: 1 6
Business Application
Server Admins
• Manual processes
Requirements
Setup Bare metal Configure
Servers Provisioning Servers
Network Adm
IT Add Users Create Update
Planning and Groups 5 VLANs Trunks
Business Outcome: Create Create UCS Configure
Define
Approvals Network Policies Service Profiles SAN Zoning
Cost Models
• Days/weeks/months to
deploy IT services
4
• High operational cost Create Create Add vFilers
Storage
IP space vFilers to Group
Management
• Consistency in CONFIGURATION across three
technology silos Compute
• Quick TIME TO MARKET for application owners
• Consistency in ACCURACY Network
Management
• Lack of features
• Possibility of Irrelevance Compute
Chassis
Blade
Blade
Eth
Eth
FC
FC
• Fragmented APIs, that eventually can be consolidated using umbrella applications, increasing the
complexity
Challenges:
Scaled Infrastructure
Manual, repetitive, error-prone tasks
Waste time & talent
Network lags behind industry automation Typo? Start from scratch
capabilities
Summary of the Amazon EC2 and Amazon RDS The Good, the OK & the Ugly of
Service Disruption in the US East Region Cloud Architecture
AWS official statement : [...] By now, it should have been drilled into
http://aws.amazon.com/message/65648/ everyone’s heads that they need to architect
“The trigger for this event was a network configuration for failure if they want guaranteed high
change. We will audit our change process and increase availability from their cloud computing efforts.
the automation to prevent this mistake from happening in
the future. However, we focus on building software and http://gigaom.com/cloud/the-good-the-alright-
services to survive failures. Much of the work that will and-the-ugly-of-cloud-architecture/
come out of this event will be to further protect the EBS
service in the face of a similar failure in the future”. (distribution of risk, careful monitoring,
analysis and reaction)
The originating cause of the disaster was the review of a
network configuration (human error).
• The automation capabilities of existing tools is not enough
• Expensive additional tools are required
• Often these tools bring silo-functionality, and they need to be integrated with
professional services
• Own custom automation development is too expensive
Save Resources Human Error Customize Innovate
Open Source Open Standards Open APIs
“ OpFlex demonstrates “We also believe Cisco’s open “Cisco OpFlex…makes it much “We‘re big believers in open
Cisco's commitment to an standards-based approach with easier for us to benefit from ACI standards and
open, application centric the ACI OpFlex protocol makes and its policy model in our open architectures…so we have
approach to cloud services..” ACI even stronger.” OpenStack deployment.” a say in how things evolve.”
Major Financial
Services Firm
Policy Architected Datacenter Enables Fast IT
Fast
IT
UCS Integrated
UCS
Compute Nexus Network
& APIC Storage
Infrastructure Solutions
2009: Industry’s First 2013: ACI for the Data 2014: Leading a New
Application Centric CenterSecurity
Fabric Market Category
Infrastructure
Traditional Element Configuration
LAN SAN
• Subject matter experts consumed by • VLAN assignments for NICs • Number of vHBAs
• VLAN tagging config for NICs • HBA WWN
manual configuration chores assignments
• FC Boot Parameters
• Remote KVM IP settings • HBA firmware
• Serial processes and multiple touches • Call Home behavior
• Remote KVM firmware
inhibit provisioning speed
• Configuration drift and maintenance •
•
Server UUID
Serial over LAN settings
• RAID settings
• Disk scrub actions
challenges •
•
Boot order
IPMI settings
• BIOS scrub actions
• BIOS firmware
• BIOS Settings
UCS Is Policy Computing (Stateless)
Stateless computing facilitates much greater scalability and flexibility
Unified
Management a
Server Policy…
Storage Policy…
Network Policy…
Virtualization Policy…
Application Profiles…
Unified, Embedded Management
Unified
Management a
Virtualization Policy…
Application Profiles…
Unified, Embedded Management
Unified
Management a
Server Policy…
LAN, SAN Config
Server Name
Virtualization Policy… UUID, MAC, WWN
Boot Information
LAN, SAN Config
Application Profiles… Firmware Policy
Provisioning & Administration Cost Reduction
Eliminate Manual
• UCS Manager embedded in all system devices
Configuration
• UCS Manager: Auto-discovery, self integrating components
Automation
Replaces Repetitive • UCS Service Profiles
Tasks • Policy based management
Extend Investment
• Seamless integration with existing tools equals no
in Existing Tools
stranded systems management investments
Network of Boxes
Needs Agility and
Time to Applications
A New OPEN Operating Model is Required
Software-Based
Network Virtualization
Lacks Scale,
Visibility, Security
More Complexity
Decreases Reliability
Disjointed Overlay
and Underlay
Increases Costs
A New OPEN Operating Model is Required
Software-Based
Network Virtualization
Lacks Scale,
Visibility, Security
More Complexity
Decreases Reliability
Disjointed Overlay
and Underlay
Increases Costs
A New OPEN Operating Model is Required
Software-Based
Network Virtualization
Lacks Scale,
Visibility, Security
Application Centric Infrastructure
More Complexity
Network Virtualization
Decreases Reliability Open Single Point of Control and
Disjointed Overlay Management (Policy)
and Underlay Application Policy and Automation
Physical and Virtual
Scale and Security
Visibility and Troubleshooting
An Innovative New Approach to Policy Application
Connectivity
Security F/W WEB APP DB
QoS OUTSIDE ADC
L4-7 Services ADC
APPLICATION
NETWORK PROFILE
1. End Point Group: A set of VMs / servers with the same policy
2. Contracts: A set of rules governing communication between groups
3. Service Chains: A set of network services between groups
Stateless Networking
Application Network Profile
SLA APIC
QoS
CONNECTIVITY SECURITY
Security APPLICATION L4..7 STORAGE AND
QOS
Network Profile
POLICY POLICIES SERVICES COMPUTE
Load
Balancing
APP PROFILE
application-centric network policy
On-Demand
Domain
Secure Cloud Automated Delivery Managers
Container
OS and
Network Compute VMs Storage Virtual Bare
VM VM
Metal
Machines
Virtualized and Bare-Metal
Policy-Driven Single Pane of Glass
Provisioning Compute
Compute and Hypervisor
End-to-End
Infrastructure Infrastructure
Automation and Automation and Network A B C
Management
Lifecycle Management Network and Services
On-Demand
Domain
Secure Cloud Automated Delivery Managers
Container
OS and
Network Compute VMs Storage Virtual Bare
VM VM
Metal
Machines
Virtualized and Bare-Metal
Policy-Driven Single Pane of Glass
Provisioning Compute
Compute and Hypervisor
End-to-End
Infrastructure
UCS Director Automation and Network A B C
Lifecycle Management Network and Services
Dynamic Capacity Management Metering & Chargeback Management & Administration UCS Director
Dynamic,
Reduced Accelerated Efficient, Security Consistency ITaaS
Complexity Deployment Agile
SaaS
Hybrid
Public Private
IT Sourcing Spectrum
Cisco Offers a Complete Portfolio to Support Your Cloud
Initiatives
CISCO CLOUD PRODUCTS AND Cisco Cloud Marketplace
SOLUTIONS for ENTERPRISE
© 2014 Cisco and/or its affiliates. All rights reserved. Cisco Confidential 34