Network Availability (2.2 & 3.3) Flashcards
Network Availability
o Measure of how well a computer network can respond to connectivity and
performance demands that are placed upon it
High Availability
▪ Availability is measured by uptime
▪ Five nines of availability (99.999%)
▪ Maximum of 5 minutes of downtime per year
Availability
o Concerned with being up and operational
Reliability
o Concerned with not dropping packets
Mean Time to Repair (MTTR)
o Measures the average time it takes to repair a network
device when it breaks
Mean Time Between Failures (MTBF)
o Measures the average time between failures of a device
Redundant Network with Single Points of Failure
▪ Link Redundancy (Multiple connections between devices)
● Internal Hardware Redundancy (Power supplies and NICs)
Redundant Network with No Single Points of Failure
▪ Link Redundancy (Multiple connections between devices)
● Redundancy of Components (Switches and Routers)
Hardware Redundancy
▪ Takes many forms
▪ Devices with two network interface cards (NICs), hard drives, or internal
power supplies
▪ Often found in strategic network devices
● Routers, Switches, Firewalls, and Servers
● Not often found in clients due to costs and administrative
overhead involved in management
Active-Active
▪ Multiple NICs are active at the same time
▪ NICs have their own MAC address
▪ Makes troubleshooting more complex
Active-Passive
▪ One NIC is active at a time
▪ Client appears to have a single MAC address
Network Interface Card Teaming
Network Interface Card Teaming
▪ Using a group of network interface cards for load balancing and failover
on a server or other device
Layer 3 Redundancy
▪ Clients are configured with a default gateway (router)
● If the default gateway goes down, they cannot leave the subnet
● Layer 3 Redundancy occurs with virtual gateways
Hot Standby Router Protocol (HSRP)
▪ Proprietary first-hop redundancy by Cisco
▪ Allows for active router and standby router
▪ Creates virtual router as the default gateway
Virtual Router Redundancy Protocol (VRRP)
▪ IETP open-standard variant of HSRP
▪ Allows for active router and standby router
▪ Creates virtual router as the default gateway
Gateway Load Balancing Protocol (GLBP)
▪ Proprietary first-hop redundancy by Cisco
▪ Focuses on load balancing over redundancy
▪ Allows for active router and standby router
▪ Creates virtual router as the default gateway
Link Aggregation Control Protocol (LACP)
▪ Achieves redundancy by having multiple links
between devices
▪ Load balancing occurs over multiple links
▪ Multiple links appear as single logical link
Link Aggregation Control Protocol (LACP)
▪ Achieves redundancy by having multiple links
between devices
▪ Load balancing occurs over multiple links
▪ Multiple links appear as single logical link
▪ Creates more than one physical path between the server and its storage
devices for better fault tolerance and performance
Design Considerations
▪ Where will redundancy be used?
● Module (or Parts) Redundancy
● Chassis Redundancy
▪ What software redundancy features are appropriate?
▪ What protocol characteristics affect design requirements?
▪ What redundancy features should be used to provide power to an
infrastructure device?
▪ What redundancy features should be used to maintain environmental
conditions?
Best Practices
est Practices
▪ Examine the technical goals
▪ Identify the budget to fund high availability features
▪ Categorize business applications into profiles
● Each requires a certain level of availability
▪ Establish performance standards for high-availability solutions
● Performance standards will drive how success is measured
▪ Define how to manage and measure the high-availability solution
● Metrics help quantify success to decision makers
Remember…
▪ Existing networks can be retrofitted, but it reduces the cost by integrating
high availability practices and technologies into your initial designs
Cold Sites
old Sites
▪ An available building that does not have any hardware or software in
place or configured
▪ While recovery is possible, it is going to be slow and time-consuming
Warm Sites
▪ An available building that already contains a lot of the equipment
▪ Restoral time is between 24 hours and seven days
Hot Sites
▪ An available building that already has the equipment and data in place
and configured
▪ Minimal downtime and with nearly identical service levels maintained
Cloud Site
▪ Allows for the creation of a recovery version of an organization’s
enterprise network in the cloud
Recovery Time Objective (RTO)
o Time and service level within which a business process
must be restored after a disaster to avoid unacceptable
consequences
o How much time did it
take to recover after the notification of a business process
disruption?
o Use either a hot site or a cloud site for low RTO situations
Recovery Point Objective (RPO)
o Interval of time during a disruption before data lost
exceeds the BCP’s maximum allowable threshold or
tolerance
Backup and Recovery
▪ Full
● Complete backup is the safest and most comprehensive; Time
consuming and costly
▪ Incremental
● Backup only data changed since last backup
▪ Differential
● Only backups data since the last full backup
▪ Snapshots
● Read-only copy of data frozen in time (VMs)
Uninterruptible Power Supply (UPS)
▪ Provides emergency power to a load when the input power source or
main power fails
▪ Great for short duration power outages (less than 15 minutes)
Power Distribution Unit (PDU)
▪ Distributes electric power, especially to racks of computers and
networking equipment located within a data center
▪ PDUs combined with a UPS or a generator can provide power during a
blackout
Generator
▪ Provides long-term power during a power outage in a region
● Takes a while to start up
● Hot and cold aisle concept
Wet Pipe System
▪ Using a sprinkler system and pipes that always contain water
Pre-Action System
▪ A detector actuation like a smoke detector and a sprinkler must be
tripped prior to water being released
Need for Quality of Service (QoS)
▪ Networks carry data, voice, and video content
▪ Convergence of media on the network requires high availability to ensure
proper delivery
▪ Optimizing the network to efficiently utilize the bandwidth to deliver
useful solutions to network users is crucial to success and cost savings
Quality of Service (QoS)
▪ Enables strategic optimization of network performance for different
types of traffic
● Identifies types of traffic needing priority
● Determines how much bandwidth required
● Efficiently uses WAN link’s bandwidth
● Identifies types of traffic to drop during network congestion
▪ For example:
● Voice (VoIP) and Video should have higher priority levels (less
latency)
Categories of QoS
▪ Delay ● Time a packet travels from source to destination ● Measured in milliseconds (ms) ▪ Jitter ● Uneven arrival of packets ● Especially harmful in VoIP ▪ Drops ● Occurs during link congestion ● Router’s interface queue overflows and causes packet loss
“Effective” Bandwidth
Pictures
“Effective” Bandwidth
Pictures
Purpose of QoS
▪ To categorize traffic, apply a policy to those traffic categories, and
prioritize them in accordance with a QoS policy
Categorization of Traffic
ategorization of Traffic
▪ Determine network performance requirements for various traffic types
(Voice, Video, Data)
▪ Categorize traffic into specific categories:
● Low delay
o Voice
o Streaming Video
● Low priority
o Web browsing
o Non-mission critical data
▪ Document your QoS policy and make it available to your users
Best Effort
● Does not truly provide QoS to that traffic ● No reordering of packets ● Uses FIFO (first in, first out) queuing
Integrated Services (IntServ or Hard QoS)
● Makes strict bandwidth
reservations
● Reserves bandwidth by signaling devices
Differentiated Services (DiffServ or Soft QoS)
● Differentiates between multiple traffic flows
● Packets are “marked”
● Routers and switches make decisions based on those markings
Methods of Categorizing Traffic
▪ Classification ▪ Marking ▪ Congestion management ▪ Congestion avoidance ▪ Policing and shaping ▪ Link efficiency
Ways of Categorizing Traffic
▪ Classification ▪ Marking ▪ Congestion management ▪ Congestion avoidance ▪ Policing and shaping ▪ Link efficiency
Classification of Traffic
▪ Traffic is placed into different categories
▪ For example, the E-mail class might contain various types of traffic
● POP3
● IMAP
● SMTP
● Exchange
▪ Classification does not alter any bits in the frame or packet
Marking of Traffic
▪ Altered bits within a frame, cell, or packet indicates handling of traffic
▪ Network tools make decisions based on markings
Congestion Management
▪ When a device receives traffic faster than it can be transmitted, it buffers
the extra traffic until bandwidth becomes available
● Called queuing
▪ Queuing algorithm empties the packets in specified sequence and
amount
▪ Queuing algorithms types
● Weighted fair queuing
● Low-latency queuing
● Weighted round-robin
Congestion Avoidance
▪ Newly arriving packets would be discarded if the device’s output queue
fills to capacity
▪ Random Early Detection (RED) is used to prevent this from occurring
● As the queue fills, the possibility of a discard increases until it
reaches 100%
● If at 100%, all traffic of that type is dropped
● RED instead drops packets from selected queues based on
defined limits
▪ If TCP traffic, it will be retransmitted
▪ If UDP, it will simply be dropped
Policing and Shaping
▪ Policing
● Typically discards packets that exceed a configured rate limit
(speed limit)
● Dropped packets result in retransmissions
● Recommended for higher-speed interfaces
▪ Shaping
● Buffers (delays) traffic exceeding configured rate
● Recommended for slower-speed interfaces
Link Efficiency: Compression
▪ Packet payload is compressed to conserve bandwidth
▪ VoIP payload can be reduced by 50%
● Payload size from 40 bytes to 20 bytes
▪ VoIP header can be reduced by 90-95%
● Uses RTP header compression (cRTP)
● Header size goes from 40 bytes to 2 to 4 bytes
▪ Utilized on slower-speed links to make most of limited bandwidth
Link Efficiency: LFI
▪ Link Fragmentation & Interleaving (LFI)
▪ Fragments large data packets and interleaves smaller data packets
between the fragments
▪ Utilized on slower-speed links to make the most of limited bandwidth
