Never Down, Never Die, Always Trade

1
Never Down, Never Die, Always Trade
A Design for an Advanced High-Availabilityand
Fault-Tolerant Architecture
Mark-Allen PerrySystems ArchitectALPHA Systems,
Switzerland
2
Agenda

• Who Needs HA-FT ?
• Defining High-Availability Fault-Tolerance
  (HA-FT)
• Why Citrix ?
• The Building Blocks of a HA-FT System
• Summary

3
Who Needs a HA-FT Architecture

• Financial institutions
• Pharmaceutical companies
• Shipping companies
• Department stores
• Fast-food companies
• Transportation companies
• just about everyone

4
Defining High-Availability

• How long a service is useable by the users
• Individual systems can accurately be measured but
  not the complete architecture

5
Defining Fault Tolerance

• The ability to maintain services during
  unscheduled events
• Level of redundancy
• Infrastructure, hardware, and network
• Infrastructure architecture
• System design
• System management
• Experience of personnel

6
Risks of Low Availability and No Fault Tolerance

• Risk to the bottom-line
• Users can't get to application
• Clients can't get to their information
• Risk to client loss
• Clients will go elsewhere to do business
• Lower increase in new business
• Risk of loss of reputation
• Unreliable service

7
Questions Decisions

• Why did we select Citrix MetaFrame?
• Need to provide a better fault-tolerant system
• Need to project the display across the network,
  anywhere in the world
• Need to protect against uncontrolled changes
• Need to protect systems in a secure environment
• Reliability

8
Building Blocks of HA-FT Systems

• Physical Infrastructure
• Power systems, UPS, A/C, system rooms
• Networking Components
• Racks, cabling, switches
• DC, DNS, DHCP
• Optimized Operating System build
• Citrix design and configuration
• Intelligent application design
• Competent personnel

9
Physical Infrastructure

• Multiple locations floors rooms
• Power Systems
• Sufficient UPS
• Air Conditioning
• designed for a loss of A/C
• System Rooms

10
Networking Infrastructure Rack Design
11
Networking Infrastructure

• Cabling - multiple
• Normal, a backup, and RIB
• Switches
• Domain Controllers
• DNS Servers
• DHCP Servers

12
Optimized Operating System Build

• Prime Directive
• Never down, never die, always trade
• Over-build hardware
• HW fault-tolerant
• Light, Clean, and Fast
• Easily imaged (staged)
• Monitored
• Upgradeable

13
Intelligent Citrix Design

• Understand Citrix
• Test different designs
• Pilot
• Idiosyncrasies
• Naming conventions
• Positioning
• Round-Robin allocation

14
Application Design Integration

• If you can build it for HA-FT, do so
• Know the application
• Study the applications process
• Determine if you can use the application to add
  an additional layer of fault-tolerance
• Installation/upgrade should be simple and
  verifiable

15
Personnel

• Often forgotten but most important
• Regular education
• Think out-of-the-box
• In an event, they will provide the first line of
  recovery

16
What Kind of HA-FT Do I Have?

• Granular Fail-over Design
• a single server
• a single PDU within a rack
• an entire rack
• a single switch
• a system room including UPS, air conditioning and
  network
• A complete site, where the DR site would handle
  all traders

17
Final Points

• Allow for expansion
• People will always find a use for Citrix
• Have sufficient Swing servers
• Have sufficient test servers
• Think
• Anticipate
• Prepare

18
Thank You! Questions