Unified Computing

I work for Cisco in their Unified Computing Advanced Services team in Europe.  All things UCS go in this page.

Unified Computing System Patterns and Implementations

I’m working to extend ITIL to have UCS specific patterns and implementations, starting with Unified Event Management.  These will be a series of collaborative articles (working with other industry vendors and consultants) that provide concrete implementations for ITIL best practice.

• Unified Event Management

ViewYonder Posts

• Cisco UCS dog food tastes nice
• Unified Computing = Unified People, Process and Technology
• Understanding and untangling the data center spaghetti
• Cisco UCS and vSphere Management: Inside, or outside?
• New certification? DCUCD from Cisco
• Indirection, Abstraction, Pools and Policies – the UCS simple approach
• Brush up on your drawing skills for UCS and vSphere
• Why does Cisco UCS memory stop at 384GB RAM?
• vSphere Resource Management: a factory artifact
• Toyota Production System for I

Experts Online

Who’s who when it comes to UCS? There are many ways to find the experts, these days, from the communities to blogs to Twitter. Here’s the list of people I recognize as UCS experts.

• Dave Alexander has a new site on Unified Computing.  Dave is a trainer around UCS.  http://www.unifiedcomputingblog.com/
• Brad Hedlund is a Cisco superstar who looks after InternetWork Expert.org. He writes about all things UCS like pricing and architecture. You can also find Brad on Twitter and LinkedIn.
• Chris (The) Hoff has
  • Interesting dialog on the Cisco UCS stuff (and a bit of detail)
  • The UFC and UCS: Cisco Is Brock Lesnar
• Rodos at rodos.hayood.org is a long time VMware blogger and was the first moderator on VIOPS. He’s an Enterprise Architect and he collects UCS musings.. He’s also on Twitter and LinkedIn.
  • Scaling Cisco Unified Computing System (UCS)
  • Nehalem Memory with Catalina
• Chad Sakac of EMC needs no introduction as probably the only VP who calls himself a Propeller Head!  His superb blog VirtualGeek has posts on UCS like
  • Cisco UCS Manager in Action
  • Interesting dialog on the Cisco UCS stuff (and a bit of detail)

• Scott Lowe with
  • Mudslinging Between Cisco and HP
  • Cisco UCS Virtualization-Optimized CNAs
• SearchStorage has a PRIMER: Cisco’s Systems Unified Computing System for storage pros
• Colin McNamara writes
  • Cisco Unified Computing System Quoting and Configuration with Netformx
  • Setting the record straight about pricing, in response to the Register article Cisco California pricing revealed

Official Cisco Stuff

Videos

• Unified Computing System Manager Revealed (Part 1)
• Unified Computing System Manager Revealed (Part 2)
• Cisco Unified Computing System Manager and Firmware Profiles
• Rack and Install the Cisco (UCS) 6120XP Chassis
• Unboxing the Cisco UCS Server Chassis
• Rack and Install the Cisco UCS 5108 Server Chassis

Documentation

• Documentation Entry Page, CCO login required
  • Release Notes for Cisco UCS Release 1.0(1e)
  • Site Preperation Guide
  • Hardware and Software Interoperability Matrix
  • Configuration Examples and Notes, like configuring DNS
  • Configuration Guides, for GUI and CLI
  • Programming Guide, for the API – note that the link on this page is a bit dodgy (no .htm suffix so browser isn’t sure how to parse that content type) – here’s the PDF.
  • Troubleshooting Guide

Whitepapers

• Scalability Study for Deploying VMware View on Cisco UCS and EMC V-Max Systems (get PDF instead). In this excellent paper:
  • One half-slot blade with 48GB RAM can easily run 128 View “Knowledge Worker” desktops, including coping with Boot Storms (e.g. blade has an unplanned restart causing all desktops to restart at the same time). There is linear scaling within the chassis (8 half-slot blades = 1,024 View desktops / chassis) and potentially to a full UCS deployment of 40 chassis / 320 blades capable of 40,960 desktops supported by EMC V-Max.
  • 149 desktops per blade, 128 or 100? On page 2, there’s a formula for vSphere HA of TRUNC(170 x (n-1) / n), which for an 8 blade chassis is 149 View guests per blade. In the vSphere Configuration Maximums document there is a limit of 100 guests per host in a HA cluster, with a maximum of 1,280 guests per cluster. If there is a limit of 800 View desktops per 8-blade chassis, then that’s a %22 reduction in ROI on the recommended 128 per blade, and a 33% reduction in the above HA formula, as measured by desktops deployed. These are tested rules of thumb, and no replacement for one’s own testing.
• Unified Computing Overview (pdf)
• Cisco UCS Manager Architecture (pdf)
• Cisco Unified Computing System Extended Memory Technology Overview

Books

• Project California: A data center virtualization server by Silvano Gai, Tommi Sali and Roger Andersson
• Data Center Networks and Fibre Channel over Ethernet (FCoE) by Silvano Gai