SlideShare a Scribd company logo
1 of 29
Copyright Avi Networks 2018
High Availability
Nathan McMahon
Product Management
nathan@avinetworks.com
Copyright Avi Networks 2018
High Availability
• Why change the HA model?
• How has the model changed?
• Specific examples of impact
Active Standby
Copyright Avi Networks 2018
Islands of Technology
LB2LB1
LB2LB1
LB2LB1
LB2LB1
LB2LB1
LB2LB1
LB2LB1
LB2LB1
LB2LB1
LB2LB1
LB2LB1
LB2LB1
LB2LB1
LB2LB1
LB2LB1
LB2LB1
App VIP Data Center LB Pair LB IP Addr
Exchange 17.234.11.10 SV1 DC1-LB07a 10.120.23.34
Exchange 219.2.40.121 Virginia DC-V-LB01 10.8.10.241
OWA 17.234.11.11 SV1 DC1-LB07a 10.120.23.34
OWA 219.2.40.127 Virginia DC-V-LB01 10.8.10.241
www 17.234.28.24 SV1 DC1-LB2 10.120.23.117
AppStack 17.234.28.25 SV1 DC1-LB1 10.120.23.120
• Active / Standby load balancer pair has limited capacity
• Manual VS placement onto a single pair of LBs
• Management complexity increases with more apps
• Hard to write automation to point to the correct LBs
Active Standby
Copyright Avi Networks 2018
Islands of Technology
Active
15%
Standby
0%
• No shared capacity pooling
• Costly overprovisioning
• Shift from proprietary hardware to software compounds these challenges
• Average utilization of traditional LBs? 6-8 %
Copyright Avi Networks 2018
What if we change the model
Active Standby
Copyright Avi Networks 2018
CONTROLDATA
Service Engines
Controllers
Separate
Control and
Data Plane
Manage as one,
not many devices
Copyright Avi Networks 2018
Bare Metal Virtualized Containers Public Cloud
CONTROLDATA
Service Engines
Controllers
MESOS
Hybrid Cloud
Both traditional and modern use cases
Automation
Highly programmable, plug-n-play
Analytics
Actionable insights key to automation
Separate
Control and
Data Plane
Manage as one,
not many devices
Copyright Avi Networks 2018
Avi Object Model
Copyright Avi Networks 2018
Avi Object Model
• Avi Controller
• Avi Service Engines
• Load Balancing Components
Virtual Service Pools Networks Servers
Copyright Avi Networks 2018
Controller HA
Copyright Avi Networks 2018
Controller Redundancy
• Controller may be deployed as a standalone, or a redundant three node cluster
• High availability uses a Zookeeper-like model of a 3 node cluster to maintain a quorum
• All Controllers are active, sharding workloads
• Management may be performed from any Controller in the cluster
Controller Cluster
3 Node Cluster
Standalone
Leader
Follower Follower
Copyright Avi Networks 2018
Single Node Failure
• No impact to data plane (the Service Engines) or management
Controller High Availability
Copyright Avi Networks 2018
Single Node Failure
• No impact to data plane (the Service Engines) or management
Two Node Failure
• The remaining Controller node will not take over as active without quorum (2 nodes)
– Mitigates split-brain issue with traditional A/A, such as if one Controller was not down but merely lost connectivity to peers
• Remaining Controller must be manually promoted to own the cluster and be active
Controller High Availability
?
?
Copyright Avi Networks 2018
Single Node Failure
• No impact to data plane (the Service Engines) or management
Two Node Failure
• The remaining Controller node will not take over as active without quorum (2 nodes)
– Mitigates split-brain issue with traditional A/A, such as if one Controller was not down but merely lost connectivity to peers
• Remaining Controller must be manually promoted to own the cluster and be active
Three Node Failure
• No impact to data plane. Service Engines continue to run in headless mode until Controllers are restored
• No configuration changes possible until Controllers are restored / redeployed
• Service Engines will buffer metrics and logs until Controllers are back. Buffer size depends on disk allocation for SEs
Controller High Availability
Copyright Avi Networks 2018
Controller Process Sharding
• All Controllers are actively working, though they may be doing different tasks
• Each virtual service is hashed to a Controller to divide the workload
• Many newer environments are built around 3 availability zones
Controller cluster sharding
workloads from four virtual services
VS1 VS2 VS2 VS4VS3
Leader Follower Follower
Copyright Avi Networks 2018
Service Engine HA
Copyright Avi Networks 2018
Templates
• SE Groups contain sizing, scaling, placement and HA properties
• A new SE will be created from the SE Group properties
• SE Group options will vary based upon the cloud / ecosystem
Folders
• An SE is always a member of the group it was created within
• Each SE group is an isolation domain
• Apps may gracefully migrate, scale, or failover across SEs in the group
• Client session data automatically replicated to other SEs in the group
– Persistence tables
– SSL session/tickets
– DataScript variables
SE Groups
100 Avi-SE-xyz
70 Avi-SE-abc
100 Avi-SE-def
SEs: 2vCPU, 2Gb
HA: Active / Active
SE Group 2
! Avi-Lab-123
! Avi-Lab-456
SEs: 1vCPU, 1Gb
HA: Active / Standby
SE Group 1
Copyright Avi Networks 2018
SE High Availability Modes
Fastest failover time
Least efficient SE utilization
Longest failover time
Most efficient SE utilization
Legacy
Active / Standby
Elastic
Active / Active
Elastic
N + M
Elastic
N + 0
Failover Steps
SE failure detection
Controller determines SE to fail over to
Controller creates new SE
Copy VS configuration to new SE
Configure vNIC on new SE
Move VIP via GARP or cloud API
Copyright Avi Networks 2018
Legacy Active/Standby
• VS is active on one SE, standby on another
• No VS scaleout support
• Primarily for default gateway / non-SNAT app support
• Fastest failover, but half of SE resources are idle
SE High Availability Modes
SE 1
Active
SE 2
Standby
Steady state
App 3
App 2
App 1
SE 1
Down
SE 2
Active
Failed SE state
App 3
App 2
App 1
High Availability Mode A / S
SE failure detection O
Controller determines SE to fail over to -
Controller creates new SE -
Copy VS configuration to new SE -
Configure vNIC on new SE -
Move VIP via GARP or cloud API O
Copyright Avi Networks 2018
Elastic Active / Active [best practice for production apps]
• All SEs are active
• VS must be scaled across at least 2 SEs
• SE failover decision pre-determined
• Session info proactively replicated to other scaled SEs
• Faster failover, potentially greater SE resource requirement
Elastic N + M [default mode]
• All SEs are active
• N = number of SEs a new VS is scaled across
• M = the buffer, or number of failures the group can sustain
• SE failover decision determined at time of failure
• Session replication done after new SE is chosen
• Slower failover, less SE resource requirement
SE High Availability Modes
SE 1 SE 2 SE 3
SE 1 SE 2 SE 3
SE 1 SE 2 SE 3 SE 4
Steady state, each SE utilized
One SE fails
New SE created to meet HA requirement
App 2
App 1
App 2
App 3
App 2
App 4
App 2
App 1
App 3
App 2
App 4
App 2
App 1
App 2
App 4 App 3
App 2
High Availability Mode A / A N + M
SE failure detection O O
Controller determines SE to fail over to O
Copy VS configuration to new SE O
Configure vNIC on new SE O
Move VIP via GARP or cloud API O O
Copyright Avi Networks 2018
SE High Availability Modes
Fastest failover time
Least efficient SE utilization
Longest failover time
Most efficient SE utilization
Legacy
Active / Standby
Elastic
Active / Active
Elastic
N + M
Elastic
N + 0
High Availability Mode A / S A / A N + M N + 0
SE failure detection O O O O
Controller determines SE to fail over to O O
Controller creates new SE O
Copy VS configuration to new SE O O
Configure vNIC on new SE O O
Move VIP via GARP or cloud API O O O O
Copyright Avi Networks 2018
SE Native Scaling
Automatically Increase Service Engine Capacity
1. Traffic is steady for a virtual service.
The primary SE ARPs for the VIP address.
2. Traffic increases beyond the capacity of a single SE.
3. Controller brings new load balancers (SEs) online.
4. The primary SE delegates some traffic to new SEs by
forwarding some connections (L2 switched) to the MAC addresses
of the other SEs.
5. Each SE takes a portion of the load.
With SNAT, servers return traffic to the source SE MAC.
SEs forward response traffic directly back to clients.
SE 1
Copyright Avi Networks 2018
Scale Service Engines via Upstream Router
• All SEs advertise the VIP to BGP via Route Health Injection
• Router hashes client flows across SEs
• ECMP mode enables scaling across 2 to 64 Service Engines
• With SNAT, servers return traffic to the source SE MAC address
• SEs send response traffic directly to clients
Failure Mitigation
• BFD may be enabled to ensure faster detection of an SE failure
• Persistence and SSL connections are mirrored to ensure a graceful
and automatic recovery in case of a router hash redistribution
• SEs will forward incorrectly hashed flows to the proper SE
SE ECMP Scaling
SE
Copyright Avi Networks 2018
SE Auto Scaling
SE 1
Scaling
• Scale Out
• Scale In
– Gracefully remove an SE from the active/active group
– Waits one minute for connections to close before scaling in
• Migrate
1. Scale out from SE1 to SE2
2. SE2 GARPs for the VIP
3. Scale in to SE2, removing SE1 from servicing the VIP
Manual Scaling
• Administrator initiated scale in, out, and migrate
• Default mode
Auto Scaling
• SE Group may be configured for manual or automatic scaling
• Avi does not [yet] recommend auto scaling
– Works for CPU above/below threshold
– Auto scale available via CLI/API
Copyright Avi Networks 2018
Scale SE Performance Up and Out
SE
SE SE
SE
SE SE
Scale up with more CPU cores
Scale out with more SEs
Copyright Avi Networks 2018
Multi Availability Zones for Public Cloud
• Public clouds such as AWS split a data center into three Availability Zones
• Each AZ is a separate IP network space
• AWS customers are expected to load balancing traffic into the three Azs
• Avi deploys an SE per AZ
• DNS is then used to distribute traffic across the three VIP addresses for an app
• The Avi Controller removes a VIP from DNS if that AZ or SE is down
• Multi AZ awareness for AWS and Azure require a DNS profile for the cloud AZ 1 AZ 2 AZ 3
www.avi.com
20.1.1.1 20.2.2.2 20.3.3.3
Traffic distribution in AWS data center with 3 AZs
Copyright Avi Networks 2018
Bare Metal Virtualized Containers Public Cloud
CONTROLDATA
MESOS
Hybrid Cloud
Both traditional and modern use cases
Automation
Highly programmable, plug-n-play
Analytics
Actionable insights key to automation
Separate
Control and
Data Plane
Manage as one,
not many devices
• Why change the HA model?
• Active/Standby is based on a physical, device-centric world
• Doesn’t scale, increases management complexity
• How has the model changed?
• NFV model, Active/Active
• Specific examples of impact
• Nearly infinite scale
• Easier management, easier to write automation
Copyright Avi Networks 2018
Next Steps
• Avi Tech Corner Webinars avinetworks.com/webinars-avi-tech-corner
• Avi Knowledge Base avinetworks.com/docs
• Avi Workshops avinetworks.com/workshops
• Virtual Lab email: education@avinetworks.com
Copyright Avi Networks 2018
Nathan McMahon
education@avinetworks.com
avinetworks.com/workshops

More Related Content

What's hot

Smart monitoring how does oracle rac manage resource, state ukoug19
Smart monitoring how does oracle rac manage resource, state ukoug19Smart monitoring how does oracle rac manage resource, state ukoug19
Smart monitoring how does oracle rac manage resource, state ukoug19Anil Nair
 
Kubernetes Introduction
Kubernetes IntroductionKubernetes Introduction
Kubernetes IntroductionPeng Xiao
 
Virtualization Vs. Containers
Virtualization Vs. ContainersVirtualization Vs. Containers
Virtualization Vs. Containersactualtechmedia
 
Kubernetes for Beginners: An Introductory Guide
Kubernetes for Beginners: An Introductory GuideKubernetes for Beginners: An Introductory Guide
Kubernetes for Beginners: An Introductory GuideBytemark
 
Percona XtraDB Cluster vs Galera Cluster vs MySQL Group Replication
Percona XtraDB Cluster vs Galera Cluster vs MySQL Group ReplicationPercona XtraDB Cluster vs Galera Cluster vs MySQL Group Replication
Percona XtraDB Cluster vs Galera Cluster vs MySQL Group ReplicationKenny Gryp
 
PostgreSQL High Availability in a Containerized World
PostgreSQL High Availability in a Containerized WorldPostgreSQL High Availability in a Containerized World
PostgreSQL High Availability in a Containerized WorldJignesh Shah
 
Running MariaDB in multiple data centers
Running MariaDB in multiple data centersRunning MariaDB in multiple data centers
Running MariaDB in multiple data centersMariaDB plc
 
Microservices Architecture & Testing Strategies
Microservices Architecture & Testing StrategiesMicroservices Architecture & Testing Strategies
Microservices Architecture & Testing StrategiesAraf Karsh Hamid
 
Microservices Part 3 Service Mesh and Kafka
Microservices Part 3 Service Mesh and KafkaMicroservices Part 3 Service Mesh and Kafka
Microservices Part 3 Service Mesh and KafkaAraf Karsh Hamid
 
Releasing Software Quickly and Reliably with AWS CodePipline
Releasing Software Quickly and Reliably with AWS CodePiplineReleasing Software Quickly and Reliably with AWS CodePipline
Releasing Software Quickly and Reliably with AWS CodePiplineAmazon Web Services
 
Docker Introduction
Docker IntroductionDocker Introduction
Docker IntroductionPeng Xiao
 
Oracle RAC Virtualized - In VMs, in Containers, On-premises, and in the Cloud
Oracle RAC Virtualized - In VMs, in Containers, On-premises, and in the CloudOracle RAC Virtualized - In VMs, in Containers, On-premises, and in the Cloud
Oracle RAC Virtualized - In VMs, in Containers, On-premises, and in the CloudMarkus Michalewicz
 
Containers orchestrators: Docker vs. Kubernetes
Containers orchestrators: Docker vs. KubernetesContainers orchestrators: Docker vs. Kubernetes
Containers orchestrators: Docker vs. KubernetesDmitry Lazarenko
 
The Zen of High Performance Messaging with NATS
The Zen of High Performance Messaging with NATS The Zen of High Performance Messaging with NATS
The Zen of High Performance Messaging with NATS NATS
 
Kubernetes Docker Container Implementation Ppt PowerPoint Presentation Slide ...
Kubernetes Docker Container Implementation Ppt PowerPoint Presentation Slide ...Kubernetes Docker Container Implementation Ppt PowerPoint Presentation Slide ...
Kubernetes Docker Container Implementation Ppt PowerPoint Presentation Slide ...SlideTeam
 

What's hot (20)

Smart monitoring how does oracle rac manage resource, state ukoug19
Smart monitoring how does oracle rac manage resource, state ukoug19Smart monitoring how does oracle rac manage resource, state ukoug19
Smart monitoring how does oracle rac manage resource, state ukoug19
 
Galera Cluster Best Practices for DBA's and DevOps Part 1
Galera Cluster Best Practices for DBA's and DevOps Part 1Galera Cluster Best Practices for DBA's and DevOps Part 1
Galera Cluster Best Practices for DBA's and DevOps Part 1
 
Kubernetes Introduction
Kubernetes IntroductionKubernetes Introduction
Kubernetes Introduction
 
Virtualization Vs. Containers
Virtualization Vs. ContainersVirtualization Vs. Containers
Virtualization Vs. Containers
 
Kubernetes for Beginners: An Introductory Guide
Kubernetes for Beginners: An Introductory GuideKubernetes for Beginners: An Introductory Guide
Kubernetes for Beginners: An Introductory Guide
 
Percona XtraDB Cluster vs Galera Cluster vs MySQL Group Replication
Percona XtraDB Cluster vs Galera Cluster vs MySQL Group ReplicationPercona XtraDB Cluster vs Galera Cluster vs MySQL Group Replication
Percona XtraDB Cluster vs Galera Cluster vs MySQL Group Replication
 
PostgreSQL High Availability in a Containerized World
PostgreSQL High Availability in a Containerized WorldPostgreSQL High Availability in a Containerized World
PostgreSQL High Availability in a Containerized World
 
Running MariaDB in multiple data centers
Running MariaDB in multiple data centersRunning MariaDB in multiple data centers
Running MariaDB in multiple data centers
 
Oracle ASM Training
Oracle ASM TrainingOracle ASM Training
Oracle ASM Training
 
Containers 101
Containers 101Containers 101
Containers 101
 
Microservices Architecture & Testing Strategies
Microservices Architecture & Testing StrategiesMicroservices Architecture & Testing Strategies
Microservices Architecture & Testing Strategies
 
Microservices Part 3 Service Mesh and Kafka
Microservices Part 3 Service Mesh and KafkaMicroservices Part 3 Service Mesh and Kafka
Microservices Part 3 Service Mesh and Kafka
 
Releasing Software Quickly and Reliably with AWS CodePipline
Releasing Software Quickly and Reliably with AWS CodePiplineReleasing Software Quickly and Reliably with AWS CodePipline
Releasing Software Quickly and Reliably with AWS CodePipline
 
Docker Kubernetes Istio
Docker Kubernetes IstioDocker Kubernetes Istio
Docker Kubernetes Istio
 
Docker Introduction
Docker IntroductionDocker Introduction
Docker Introduction
 
Kali Linux Installation - VMware
Kali Linux Installation - VMwareKali Linux Installation - VMware
Kali Linux Installation - VMware
 
Oracle RAC Virtualized - In VMs, in Containers, On-premises, and in the Cloud
Oracle RAC Virtualized - In VMs, in Containers, On-premises, and in the CloudOracle RAC Virtualized - In VMs, in Containers, On-premises, and in the Cloud
Oracle RAC Virtualized - In VMs, in Containers, On-premises, and in the Cloud
 
Containers orchestrators: Docker vs. Kubernetes
Containers orchestrators: Docker vs. KubernetesContainers orchestrators: Docker vs. Kubernetes
Containers orchestrators: Docker vs. Kubernetes
 
The Zen of High Performance Messaging with NATS
The Zen of High Performance Messaging with NATS The Zen of High Performance Messaging with NATS
The Zen of High Performance Messaging with NATS
 
Kubernetes Docker Container Implementation Ppt PowerPoint Presentation Slide ...
Kubernetes Docker Container Implementation Ppt PowerPoint Presentation Slide ...Kubernetes Docker Container Implementation Ppt PowerPoint Presentation Slide ...
Kubernetes Docker Container Implementation Ppt PowerPoint Presentation Slide ...
 

Similar to Design Best Practices for High Availability in Load Balancing

OpenStack Summit Fall 2018: LBaaS
OpenStack Summit Fall 2018: LBaaSOpenStack Summit Fall 2018: LBaaS
OpenStack Summit Fall 2018: LBaaSPraveen Yalagandula
 
More Nines for Your Dimes: Improving Availability and Lowering Costs using Au...
More Nines for Your Dimes: Improving Availability and Lowering Costs using Au...More Nines for Your Dimes: Improving Availability and Lowering Costs using Au...
More Nines for Your Dimes: Improving Availability and Lowering Costs using Au...Amazon Web Services
 
Cisco Connect 2018 Singapore - Easing the Transition
Cisco Connect 2018 Singapore - Easing the Transition Cisco Connect 2018 Singapore - Easing the Transition
Cisco Connect 2018 Singapore - Easing the Transition NetworkCollaborators
 
Netherlands Tech Tour 02 - MySQL Fabric
Netherlands Tech Tour 02 -   MySQL FabricNetherlands Tech Tour 02 -   MySQL Fabric
Netherlands Tech Tour 02 - MySQL FabricMark Swarbrick
 
More Nines for Your Dimes: Improving Availability and Lowering Costs using Au...
More Nines for Your Dimes: Improving Availability and Lowering Costs using Au...More Nines for Your Dimes: Improving Availability and Lowering Costs using Au...
More Nines for Your Dimes: Improving Availability and Lowering Costs using Au...Amazon Web Services
 
Understanding network and service virtualization
Understanding network and service virtualizationUnderstanding network and service virtualization
Understanding network and service virtualizationSDN Hub
 
The Need for Complex Analytics from Forwarding Pipelines
The Need for Complex Analytics from Forwarding Pipelines The Need for Complex Analytics from Forwarding Pipelines
The Need for Complex Analytics from Forwarding Pipelines Netronome
 
Make 2016 your year of SMACK talk
Make 2016 your year of SMACK talkMake 2016 your year of SMACK talk
Make 2016 your year of SMACK talkDataStax Academy
 
Simplifying Hyper-V Management for VMware Administrators
Simplifying Hyper-V Management for VMware AdministratorsSimplifying Hyper-V Management for VMware Administrators
Simplifying Hyper-V Management for VMware Administrators5nine
 
Transforming Legacy Applications Into Dynamically Scalable Web Services
Transforming Legacy Applications Into Dynamically Scalable Web ServicesTransforming Legacy Applications Into Dynamically Scalable Web Services
Transforming Legacy Applications Into Dynamically Scalable Web ServicesAdam Takvam
 
HTTP/2 Comes to Java: Servlet 4.0 and what it means for the Java/Jakarta EE e...
HTTP/2 Comes to Java: Servlet 4.0 and what it means for the Java/Jakarta EE e...HTTP/2 Comes to Java: Servlet 4.0 and what it means for the Java/Jakarta EE e...
HTTP/2 Comes to Java: Servlet 4.0 and what it means for the Java/Jakarta EE e...Edward Burns
 
MariaDB High Availability Webinar
MariaDB High Availability WebinarMariaDB High Availability Webinar
MariaDB High Availability WebinarMariaDB plc
 
Microservices OSGi-running-with-apache-karaf
Microservices OSGi-running-with-apache-karafMicroservices OSGi-running-with-apache-karaf
Microservices OSGi-running-with-apache-karafAchim Nierbeck
 
VMworld 2014: Extreme Performance Series
VMworld 2014: Extreme Performance Series VMworld 2014: Extreme Performance Series
VMworld 2014: Extreme Performance Series VMworld
 
Best Practice for Achieving High Availability in MariaDB
Best Practice for Achieving High Availability in MariaDBBest Practice for Achieving High Availability in MariaDB
Best Practice for Achieving High Availability in MariaDBMariaDB plc
 
Virt july-2013-meetup
Virt july-2013-meetupVirt july-2013-meetup
Virt july-2013-meetupnvirters
 
CloudStack challenges for China customers
CloudStack challenges for China customersCloudStack challenges for China customers
CloudStack challenges for China customersgavin_lee
 

Similar to Design Best Practices for High Availability in Load Balancing (20)

OpenStack Summit Fall 2018: LBaaS
OpenStack Summit Fall 2018: LBaaSOpenStack Summit Fall 2018: LBaaS
OpenStack Summit Fall 2018: LBaaS
 
Neutron scaling
Neutron scalingNeutron scaling
Neutron scaling
 
More Nines for Your Dimes: Improving Availability and Lowering Costs using Au...
More Nines for Your Dimes: Improving Availability and Lowering Costs using Au...More Nines for Your Dimes: Improving Availability and Lowering Costs using Au...
More Nines for Your Dimes: Improving Availability and Lowering Costs using Au...
 
Cisco Connect 2018 Singapore - Easing the Transition
Cisco Connect 2018 Singapore - Easing the Transition Cisco Connect 2018 Singapore - Easing the Transition
Cisco Connect 2018 Singapore - Easing the Transition
 
Netherlands Tech Tour 02 - MySQL Fabric
Netherlands Tech Tour 02 -   MySQL FabricNetherlands Tech Tour 02 -   MySQL Fabric
Netherlands Tech Tour 02 - MySQL Fabric
 
More Nines for Your Dimes: Improving Availability and Lowering Costs using Au...
More Nines for Your Dimes: Improving Availability and Lowering Costs using Au...More Nines for Your Dimes: Improving Availability and Lowering Costs using Au...
More Nines for Your Dimes: Improving Availability and Lowering Costs using Au...
 
Understanding network and service virtualization
Understanding network and service virtualizationUnderstanding network and service virtualization
Understanding network and service virtualization
 
The Need for Complex Analytics from Forwarding Pipelines
The Need for Complex Analytics from Forwarding Pipelines The Need for Complex Analytics from Forwarding Pipelines
The Need for Complex Analytics from Forwarding Pipelines
 
Symantec SDN Deployment
Symantec SDN DeploymentSymantec SDN Deployment
Symantec SDN Deployment
 
Make 2016 your year of SMACK talk
Make 2016 your year of SMACK talkMake 2016 your year of SMACK talk
Make 2016 your year of SMACK talk
 
Simplifying Hyper-V Management for VMware Administrators
Simplifying Hyper-V Management for VMware AdministratorsSimplifying Hyper-V Management for VMware Administrators
Simplifying Hyper-V Management for VMware Administrators
 
Transforming Legacy Applications Into Dynamically Scalable Web Services
Transforming Legacy Applications Into Dynamically Scalable Web ServicesTransforming Legacy Applications Into Dynamically Scalable Web Services
Transforming Legacy Applications Into Dynamically Scalable Web Services
 
HTTP/2 Comes to Java: Servlet 4.0 and what it means for the Java/Jakarta EE e...
HTTP/2 Comes to Java: Servlet 4.0 and what it means for the Java/Jakarta EE e...HTTP/2 Comes to Java: Servlet 4.0 and what it means for the Java/Jakarta EE e...
HTTP/2 Comes to Java: Servlet 4.0 and what it means for the Java/Jakarta EE e...
 
MariaDB High Availability Webinar
MariaDB High Availability WebinarMariaDB High Availability Webinar
MariaDB High Availability Webinar
 
Microservices OSGi-running-with-apache-karaf
Microservices OSGi-running-with-apache-karafMicroservices OSGi-running-with-apache-karaf
Microservices OSGi-running-with-apache-karaf
 
VMworld 2014: Extreme Performance Series
VMworld 2014: Extreme Performance Series VMworld 2014: Extreme Performance Series
VMworld 2014: Extreme Performance Series
 
SDN in the Public Cloud: Windows Azure
SDN in the Public Cloud: Windows AzureSDN in the Public Cloud: Windows Azure
SDN in the Public Cloud: Windows Azure
 
Best Practice for Achieving High Availability in MariaDB
Best Practice for Achieving High Availability in MariaDBBest Practice for Achieving High Availability in MariaDB
Best Practice for Achieving High Availability in MariaDB
 
Virt july-2013-meetup
Virt july-2013-meetupVirt july-2013-meetup
Virt july-2013-meetup
 
CloudStack challenges for China customers
CloudStack challenges for China customersCloudStack challenges for China customers
CloudStack challenges for China customers
 

More from Avi Networks

DR On Demand At Fraction of the Cost (1).pptx
DR On Demand At Fraction of the Cost (1).pptxDR On Demand At Fraction of the Cost (1).pptx
DR On Demand At Fraction of the Cost (1).pptxAvi Networks
 
Cloud_controllers_public_webinar_aug31_v1.pptx
Cloud_controllers_public_webinar_aug31_v1.pptxCloud_controllers_public_webinar_aug31_v1.pptx
Cloud_controllers_public_webinar_aug31_v1.pptxAvi Networks
 
Top 4 Reasons to Migrate From NSX Load Balancing to NSX Advanced Load Balancer
Top 4 Reasons to Migrate From NSX Load Balancing to NSX Advanced Load BalancerTop 4 Reasons to Migrate From NSX Load Balancing to NSX Advanced Load Balancer
Top 4 Reasons to Migrate From NSX Load Balancing to NSX Advanced Load BalancerAvi Networks
 
23.06.15 NSX ALB and vCD integration deepdive_webinar0615.pptx
23.06.15 NSX ALB and vCD integration deepdive_webinar0615.pptx23.06.15 NSX ALB and vCD integration deepdive_webinar0615.pptx
23.06.15 NSX ALB and vCD integration deepdive_webinar0615.pptxAvi Networks
 
Enterprises-Have-Replaced-12000-ADCs-See-Why.pptx
Enterprises-Have-Replaced-12000-ADCs-See-Why.pptxEnterprises-Have-Replaced-12000-ADCs-See-Why.pptx
Enterprises-Have-Replaced-12000-ADCs-See-Why.pptxAvi Networks
 
One And Done Multi-Cloud Load Balancing Done Right.pptx
One And Done Multi-Cloud Load Balancing Done Right.pptxOne And Done Multi-Cloud Load Balancing Done Right.pptx
One And Done Multi-Cloud Load Balancing Done Right.pptxAvi Networks
 
Virtualize Application Security Today - Hardware is No Longer Needed.pptx
 Virtualize Application Security Today - Hardware is No Longer Needed.pptx Virtualize Application Security Today - Hardware is No Longer Needed.pptx
Virtualize Application Security Today - Hardware is No Longer Needed.pptxAvi Networks
 
Deploying Elastic Self-Service Load Balancing
Deploying Elastic Self-Service Load BalancingDeploying Elastic Self-Service Load Balancing
Deploying Elastic Self-Service Load BalancingAvi Networks
 
NSX_Advanced_Load_Balancer_Solution_with_Oracle.pptx
NSX_Advanced_Load_Balancer_Solution_with_Oracle.pptxNSX_Advanced_Load_Balancer_Solution_with_Oracle.pptx
NSX_Advanced_Load_Balancer_Solution_with_Oracle.pptxAvi Networks
 
Delivering Turnkey Load Balancing in VMware Cloud with Day 0 Automation
Delivering Turnkey Load Balancing in VMware Cloud with Day 0 Automation Delivering Turnkey Load Balancing in VMware Cloud with Day 0 Automation
Delivering Turnkey Load Balancing in VMware Cloud with Day 0 Automation Avi Networks
 
Bringing SaaS Simplicity to Proactive Support & Live Threat Updates
Bringing SaaS Simplicity to Proactive Support & Live Threat UpdatesBringing SaaS Simplicity to Proactive Support & Live Threat Updates
Bringing SaaS Simplicity to Proactive Support & Live Threat UpdatesAvi Networks
 
Working From Anywhere​ with​ Advanced Load Balancing​ and ​ VMware Horizon VDI
Working From Anywhere​ with​ Advanced Load Balancing​ and ​ VMware Horizon VDI Working From Anywhere​ with​ Advanced Load Balancing​ and ​ VMware Horizon VDI
Working From Anywhere​ with​ Advanced Load Balancing​ and ​ VMware Horizon VDI Avi Networks
 
Deploying Elastic, Self-Service Load Balancing for VMware NSX-T
Deploying Elastic, Self-Service Load Balancing for VMware NSX-TDeploying Elastic, Self-Service Load Balancing for VMware NSX-T
Deploying Elastic, Self-Service Load Balancing for VMware NSX-TAvi Networks
 
Avi v20.1 — What’s New in Scalable, Multi-Cloud Load Balancing
Avi v20.1 — What’s New in Scalable, Multi-Cloud Load BalancingAvi v20.1 — What’s New in Scalable, Multi-Cloud Load Balancing
Avi v20.1 — What’s New in Scalable, Multi-Cloud Load BalancingAvi Networks
 
Enterprise-Grade Load Balancing for VMware Cloud on AWS (VMC)
Enterprise-Grade Load Balancing for VMware Cloud on AWS (VMC)Enterprise-Grade Load Balancing for VMware Cloud on AWS (VMC)
Enterprise-Grade Load Balancing for VMware Cloud on AWS (VMC)Avi Networks
 
Multi Cloud Load Balancing 101 and Hands On Lab
Multi Cloud Load Balancing 101 and Hands On LabMulti Cloud Load Balancing 101 and Hands On Lab
Multi Cloud Load Balancing 101 and Hands On LabAvi Networks
 
Deliver Modern Applications with an Elastic Load Balancing Fabric Powered by ...
Deliver Modern Applications with an Elastic Load Balancing Fabric Powered by ...Deliver Modern Applications with an Elastic Load Balancing Fabric Powered by ...
Deliver Modern Applications with an Elastic Load Balancing Fabric Powered by ...Avi Networks
 
Multi Cloud Load balancing 101 and Hands-on Lab
Multi Cloud Load balancing 101 and Hands-on LabMulti Cloud Load balancing 101 and Hands-on Lab
Multi Cloud Load balancing 101 and Hands-on LabAvi Networks
 
Multi-Cloud Load Balancing 101 and Hands-On Lab
Multi-Cloud Load Balancing 101 and Hands-On LabMulti-Cloud Load Balancing 101 and Hands-On Lab
Multi-Cloud Load Balancing 101 and Hands-On LabAvi Networks
 

More from Avi Networks (20)

DR On Demand At Fraction of the Cost (1).pptx
DR On Demand At Fraction of the Cost (1).pptxDR On Demand At Fraction of the Cost (1).pptx
DR On Demand At Fraction of the Cost (1).pptx
 
Cloud_controllers_public_webinar_aug31_v1.pptx
Cloud_controllers_public_webinar_aug31_v1.pptxCloud_controllers_public_webinar_aug31_v1.pptx
Cloud_controllers_public_webinar_aug31_v1.pptx
 
Top 4 Reasons to Migrate From NSX Load Balancing to NSX Advanced Load Balancer
Top 4 Reasons to Migrate From NSX Load Balancing to NSX Advanced Load BalancerTop 4 Reasons to Migrate From NSX Load Balancing to NSX Advanced Load Balancer
Top 4 Reasons to Migrate From NSX Load Balancing to NSX Advanced Load Balancer
 
23.06.15 NSX ALB and vCD integration deepdive_webinar0615.pptx
23.06.15 NSX ALB and vCD integration deepdive_webinar0615.pptx23.06.15 NSX ALB and vCD integration deepdive_webinar0615.pptx
23.06.15 NSX ALB and vCD integration deepdive_webinar0615.pptx
 
Enterprises-Have-Replaced-12000-ADCs-See-Why.pptx
Enterprises-Have-Replaced-12000-ADCs-See-Why.pptxEnterprises-Have-Replaced-12000-ADCs-See-Why.pptx
Enterprises-Have-Replaced-12000-ADCs-See-Why.pptx
 
One And Done Multi-Cloud Load Balancing Done Right.pptx
One And Done Multi-Cloud Load Balancing Done Right.pptxOne And Done Multi-Cloud Load Balancing Done Right.pptx
One And Done Multi-Cloud Load Balancing Done Right.pptx
 
Virtualize Application Security Today - Hardware is No Longer Needed.pptx
 Virtualize Application Security Today - Hardware is No Longer Needed.pptx Virtualize Application Security Today - Hardware is No Longer Needed.pptx
Virtualize Application Security Today - Hardware is No Longer Needed.pptx
 
Deploying Elastic Self-Service Load Balancing
Deploying Elastic Self-Service Load BalancingDeploying Elastic Self-Service Load Balancing
Deploying Elastic Self-Service Load Balancing
 
NSX_Advanced_Load_Balancer_Solution_with_Oracle.pptx
NSX_Advanced_Load_Balancer_Solution_with_Oracle.pptxNSX_Advanced_Load_Balancer_Solution_with_Oracle.pptx
NSX_Advanced_Load_Balancer_Solution_with_Oracle.pptx
 
Delivering Turnkey Load Balancing in VMware Cloud with Day 0 Automation
Delivering Turnkey Load Balancing in VMware Cloud with Day 0 Automation Delivering Turnkey Load Balancing in VMware Cloud with Day 0 Automation
Delivering Turnkey Load Balancing in VMware Cloud with Day 0 Automation
 
Bringing SaaS Simplicity to Proactive Support & Live Threat Updates
Bringing SaaS Simplicity to Proactive Support & Live Threat UpdatesBringing SaaS Simplicity to Proactive Support & Live Threat Updates
Bringing SaaS Simplicity to Proactive Support & Live Threat Updates
 
Avi workshop-101
Avi workshop-101Avi workshop-101
Avi workshop-101
 
Working From Anywhere​ with​ Advanced Load Balancing​ and ​ VMware Horizon VDI
Working From Anywhere​ with​ Advanced Load Balancing​ and ​ VMware Horizon VDI Working From Anywhere​ with​ Advanced Load Balancing​ and ​ VMware Horizon VDI
Working From Anywhere​ with​ Advanced Load Balancing​ and ​ VMware Horizon VDI
 
Deploying Elastic, Self-Service Load Balancing for VMware NSX-T
Deploying Elastic, Self-Service Load Balancing for VMware NSX-TDeploying Elastic, Self-Service Load Balancing for VMware NSX-T
Deploying Elastic, Self-Service Load Balancing for VMware NSX-T
 
Avi v20.1 — What’s New in Scalable, Multi-Cloud Load Balancing
Avi v20.1 — What’s New in Scalable, Multi-Cloud Load BalancingAvi v20.1 — What’s New in Scalable, Multi-Cloud Load Balancing
Avi v20.1 — What’s New in Scalable, Multi-Cloud Load Balancing
 
Enterprise-Grade Load Balancing for VMware Cloud on AWS (VMC)
Enterprise-Grade Load Balancing for VMware Cloud on AWS (VMC)Enterprise-Grade Load Balancing for VMware Cloud on AWS (VMC)
Enterprise-Grade Load Balancing for VMware Cloud on AWS (VMC)
 
Multi Cloud Load Balancing 101 and Hands On Lab
Multi Cloud Load Balancing 101 and Hands On LabMulti Cloud Load Balancing 101 and Hands On Lab
Multi Cloud Load Balancing 101 and Hands On Lab
 
Deliver Modern Applications with an Elastic Load Balancing Fabric Powered by ...
Deliver Modern Applications with an Elastic Load Balancing Fabric Powered by ...Deliver Modern Applications with an Elastic Load Balancing Fabric Powered by ...
Deliver Modern Applications with an Elastic Load Balancing Fabric Powered by ...
 
Multi Cloud Load balancing 101 and Hands-on Lab
Multi Cloud Load balancing 101 and Hands-on LabMulti Cloud Load balancing 101 and Hands-on Lab
Multi Cloud Load balancing 101 and Hands-on Lab
 
Multi-Cloud Load Balancing 101 and Hands-On Lab
Multi-Cloud Load Balancing 101 and Hands-On LabMulti-Cloud Load Balancing 101 and Hands-On Lab
Multi-Cloud Load Balancing 101 and Hands-On Lab
 

Recently uploaded

AWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of TerraformAWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of TerraformAndrey Devyatkin
 
Boost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdfBoost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdfsudhanshuwaghmare1
 
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...Jeffrey Haguewood
 
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost SavingRepurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost SavingEdi Saputra
 
Apidays Singapore 2024 - Modernizing Securities Finance by Madhu Subbu
Apidays Singapore 2024 - Modernizing Securities Finance by Madhu SubbuApidays Singapore 2024 - Modernizing Securities Finance by Madhu Subbu
Apidays Singapore 2024 - Modernizing Securities Finance by Madhu Subbuapidays
 
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers:  A Deep Dive into Serverless Spatial Data and FMECloud Frontiers:  A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FMESafe Software
 
Powerful Google developer tools for immediate impact! (2023-24 C)
Powerful Google developer tools for immediate impact! (2023-24 C)Powerful Google developer tools for immediate impact! (2023-24 C)
Powerful Google developer tools for immediate impact! (2023-24 C)wesley chun
 
Artificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : UncertaintyArtificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : UncertaintyKhushali Kathiriya
 
Ransomware_Q4_2023. The report. [EN].pdf
Ransomware_Q4_2023. The report. [EN].pdfRansomware_Q4_2023. The report. [EN].pdf
Ransomware_Q4_2023. The report. [EN].pdfOverkill Security
 
Strategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherStrategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherRemote DBA Services
 
MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024MIND CTI
 
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...apidays
 
Strategize a Smooth Tenant-to-tenant Migration and Copilot Takeoff
Strategize a Smooth Tenant-to-tenant Migration and Copilot TakeoffStrategize a Smooth Tenant-to-tenant Migration and Copilot Takeoff
Strategize a Smooth Tenant-to-tenant Migration and Copilot Takeoffsammart93
 
Apidays Singapore 2024 - Scalable LLM APIs for AI and Generative AI Applicati...
Apidays Singapore 2024 - Scalable LLM APIs for AI and Generative AI Applicati...Apidays Singapore 2024 - Scalable LLM APIs for AI and Generative AI Applicati...
Apidays Singapore 2024 - Scalable LLM APIs for AI and Generative AI Applicati...apidays
 
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, AdobeApidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobeapidays
 
ICT role in 21st century education and its challenges
ICT role in 21st century education and its challengesICT role in 21st century education and its challenges
ICT role in 21st century education and its challengesrafiqahmad00786416
 
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Miguel Araújo
 
Exploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone ProcessorsExploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone Processorsdebabhi2
 
Apidays New York 2024 - Accelerating FinTech Innovation by Vasa Krishnan, Fin...
Apidays New York 2024 - Accelerating FinTech Innovation by Vasa Krishnan, Fin...Apidays New York 2024 - Accelerating FinTech Innovation by Vasa Krishnan, Fin...
Apidays New York 2024 - Accelerating FinTech Innovation by Vasa Krishnan, Fin...apidays
 

Recently uploaded (20)

AWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of TerraformAWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of Terraform
 
Boost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdfBoost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdf
 
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...
Web Form Automation for Bonterra Impact Management (fka Social Solutions Apri...
 
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost SavingRepurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
 
Apidays Singapore 2024 - Modernizing Securities Finance by Madhu Subbu
Apidays Singapore 2024 - Modernizing Securities Finance by Madhu SubbuApidays Singapore 2024 - Modernizing Securities Finance by Madhu Subbu
Apidays Singapore 2024 - Modernizing Securities Finance by Madhu Subbu
 
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers:  A Deep Dive into Serverless Spatial Data and FMECloud Frontiers:  A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
 
Powerful Google developer tools for immediate impact! (2023-24 C)
Powerful Google developer tools for immediate impact! (2023-24 C)Powerful Google developer tools for immediate impact! (2023-24 C)
Powerful Google developer tools for immediate impact! (2023-24 C)
 
Artificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : UncertaintyArtificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : Uncertainty
 
Ransomware_Q4_2023. The report. [EN].pdf
Ransomware_Q4_2023. The report. [EN].pdfRansomware_Q4_2023. The report. [EN].pdf
Ransomware_Q4_2023. The report. [EN].pdf
 
Strategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherStrategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a Fresher
 
MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024
 
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
Apidays New York 2024 - The Good, the Bad and the Governed by David O'Neill, ...
 
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
+971581248768>> SAFE AND ORIGINAL ABORTION PILLS FOR SALE IN DUBAI AND ABUDHA...
 
Strategize a Smooth Tenant-to-tenant Migration and Copilot Takeoff
Strategize a Smooth Tenant-to-tenant Migration and Copilot TakeoffStrategize a Smooth Tenant-to-tenant Migration and Copilot Takeoff
Strategize a Smooth Tenant-to-tenant Migration and Copilot Takeoff
 
Apidays Singapore 2024 - Scalable LLM APIs for AI and Generative AI Applicati...
Apidays Singapore 2024 - Scalable LLM APIs for AI and Generative AI Applicati...Apidays Singapore 2024 - Scalable LLM APIs for AI and Generative AI Applicati...
Apidays Singapore 2024 - Scalable LLM APIs for AI and Generative AI Applicati...
 
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, AdobeApidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
 
ICT role in 21st century education and its challenges
ICT role in 21st century education and its challengesICT role in 21st century education and its challenges
ICT role in 21st century education and its challenges
 
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
 
Exploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone ProcessorsExploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone Processors
 
Apidays New York 2024 - Accelerating FinTech Innovation by Vasa Krishnan, Fin...
Apidays New York 2024 - Accelerating FinTech Innovation by Vasa Krishnan, Fin...Apidays New York 2024 - Accelerating FinTech Innovation by Vasa Krishnan, Fin...
Apidays New York 2024 - Accelerating FinTech Innovation by Vasa Krishnan, Fin...
 

Design Best Practices for High Availability in Load Balancing

  • 1. Copyright Avi Networks 2018 High Availability Nathan McMahon Product Management nathan@avinetworks.com
  • 2. Copyright Avi Networks 2018 High Availability • Why change the HA model? • How has the model changed? • Specific examples of impact Active Standby
  • 3. Copyright Avi Networks 2018 Islands of Technology LB2LB1 LB2LB1 LB2LB1 LB2LB1 LB2LB1 LB2LB1 LB2LB1 LB2LB1 LB2LB1 LB2LB1 LB2LB1 LB2LB1 LB2LB1 LB2LB1 LB2LB1 LB2LB1 App VIP Data Center LB Pair LB IP Addr Exchange 17.234.11.10 SV1 DC1-LB07a 10.120.23.34 Exchange 219.2.40.121 Virginia DC-V-LB01 10.8.10.241 OWA 17.234.11.11 SV1 DC1-LB07a 10.120.23.34 OWA 219.2.40.127 Virginia DC-V-LB01 10.8.10.241 www 17.234.28.24 SV1 DC1-LB2 10.120.23.117 AppStack 17.234.28.25 SV1 DC1-LB1 10.120.23.120 • Active / Standby load balancer pair has limited capacity • Manual VS placement onto a single pair of LBs • Management complexity increases with more apps • Hard to write automation to point to the correct LBs Active Standby
  • 4. Copyright Avi Networks 2018 Islands of Technology Active 15% Standby 0% • No shared capacity pooling • Costly overprovisioning • Shift from proprietary hardware to software compounds these challenges • Average utilization of traditional LBs? 6-8 %
  • 5. Copyright Avi Networks 2018 What if we change the model Active Standby
  • 6. Copyright Avi Networks 2018 CONTROLDATA Service Engines Controllers Separate Control and Data Plane Manage as one, not many devices
  • 7. Copyright Avi Networks 2018 Bare Metal Virtualized Containers Public Cloud CONTROLDATA Service Engines Controllers MESOS Hybrid Cloud Both traditional and modern use cases Automation Highly programmable, plug-n-play Analytics Actionable insights key to automation Separate Control and Data Plane Manage as one, not many devices
  • 8. Copyright Avi Networks 2018 Avi Object Model
  • 9. Copyright Avi Networks 2018 Avi Object Model • Avi Controller • Avi Service Engines • Load Balancing Components Virtual Service Pools Networks Servers
  • 10. Copyright Avi Networks 2018 Controller HA
  • 11. Copyright Avi Networks 2018 Controller Redundancy • Controller may be deployed as a standalone, or a redundant three node cluster • High availability uses a Zookeeper-like model of a 3 node cluster to maintain a quorum • All Controllers are active, sharding workloads • Management may be performed from any Controller in the cluster Controller Cluster 3 Node Cluster Standalone Leader Follower Follower
  • 12. Copyright Avi Networks 2018 Single Node Failure • No impact to data plane (the Service Engines) or management Controller High Availability
  • 13. Copyright Avi Networks 2018 Single Node Failure • No impact to data plane (the Service Engines) or management Two Node Failure • The remaining Controller node will not take over as active without quorum (2 nodes) – Mitigates split-brain issue with traditional A/A, such as if one Controller was not down but merely lost connectivity to peers • Remaining Controller must be manually promoted to own the cluster and be active Controller High Availability ? ?
  • 14. Copyright Avi Networks 2018 Single Node Failure • No impact to data plane (the Service Engines) or management Two Node Failure • The remaining Controller node will not take over as active without quorum (2 nodes) – Mitigates split-brain issue with traditional A/A, such as if one Controller was not down but merely lost connectivity to peers • Remaining Controller must be manually promoted to own the cluster and be active Three Node Failure • No impact to data plane. Service Engines continue to run in headless mode until Controllers are restored • No configuration changes possible until Controllers are restored / redeployed • Service Engines will buffer metrics and logs until Controllers are back. Buffer size depends on disk allocation for SEs Controller High Availability
  • 15. Copyright Avi Networks 2018 Controller Process Sharding • All Controllers are actively working, though they may be doing different tasks • Each virtual service is hashed to a Controller to divide the workload • Many newer environments are built around 3 availability zones Controller cluster sharding workloads from four virtual services VS1 VS2 VS2 VS4VS3 Leader Follower Follower
  • 16. Copyright Avi Networks 2018 Service Engine HA
  • 17. Copyright Avi Networks 2018 Templates • SE Groups contain sizing, scaling, placement and HA properties • A new SE will be created from the SE Group properties • SE Group options will vary based upon the cloud / ecosystem Folders • An SE is always a member of the group it was created within • Each SE group is an isolation domain • Apps may gracefully migrate, scale, or failover across SEs in the group • Client session data automatically replicated to other SEs in the group – Persistence tables – SSL session/tickets – DataScript variables SE Groups 100 Avi-SE-xyz 70 Avi-SE-abc 100 Avi-SE-def SEs: 2vCPU, 2Gb HA: Active / Active SE Group 2 ! Avi-Lab-123 ! Avi-Lab-456 SEs: 1vCPU, 1Gb HA: Active / Standby SE Group 1
  • 18. Copyright Avi Networks 2018 SE High Availability Modes Fastest failover time Least efficient SE utilization Longest failover time Most efficient SE utilization Legacy Active / Standby Elastic Active / Active Elastic N + M Elastic N + 0 Failover Steps SE failure detection Controller determines SE to fail over to Controller creates new SE Copy VS configuration to new SE Configure vNIC on new SE Move VIP via GARP or cloud API
  • 19. Copyright Avi Networks 2018 Legacy Active/Standby • VS is active on one SE, standby on another • No VS scaleout support • Primarily for default gateway / non-SNAT app support • Fastest failover, but half of SE resources are idle SE High Availability Modes SE 1 Active SE 2 Standby Steady state App 3 App 2 App 1 SE 1 Down SE 2 Active Failed SE state App 3 App 2 App 1 High Availability Mode A / S SE failure detection O Controller determines SE to fail over to - Controller creates new SE - Copy VS configuration to new SE - Configure vNIC on new SE - Move VIP via GARP or cloud API O
  • 20. Copyright Avi Networks 2018 Elastic Active / Active [best practice for production apps] • All SEs are active • VS must be scaled across at least 2 SEs • SE failover decision pre-determined • Session info proactively replicated to other scaled SEs • Faster failover, potentially greater SE resource requirement Elastic N + M [default mode] • All SEs are active • N = number of SEs a new VS is scaled across • M = the buffer, or number of failures the group can sustain • SE failover decision determined at time of failure • Session replication done after new SE is chosen • Slower failover, less SE resource requirement SE High Availability Modes SE 1 SE 2 SE 3 SE 1 SE 2 SE 3 SE 1 SE 2 SE 3 SE 4 Steady state, each SE utilized One SE fails New SE created to meet HA requirement App 2 App 1 App 2 App 3 App 2 App 4 App 2 App 1 App 3 App 2 App 4 App 2 App 1 App 2 App 4 App 3 App 2 High Availability Mode A / A N + M SE failure detection O O Controller determines SE to fail over to O Copy VS configuration to new SE O Configure vNIC on new SE O Move VIP via GARP or cloud API O O
  • 21. Copyright Avi Networks 2018 SE High Availability Modes Fastest failover time Least efficient SE utilization Longest failover time Most efficient SE utilization Legacy Active / Standby Elastic Active / Active Elastic N + M Elastic N + 0 High Availability Mode A / S A / A N + M N + 0 SE failure detection O O O O Controller determines SE to fail over to O O Controller creates new SE O Copy VS configuration to new SE O O Configure vNIC on new SE O O Move VIP via GARP or cloud API O O O O
  • 22. Copyright Avi Networks 2018 SE Native Scaling Automatically Increase Service Engine Capacity 1. Traffic is steady for a virtual service. The primary SE ARPs for the VIP address. 2. Traffic increases beyond the capacity of a single SE. 3. Controller brings new load balancers (SEs) online. 4. The primary SE delegates some traffic to new SEs by forwarding some connections (L2 switched) to the MAC addresses of the other SEs. 5. Each SE takes a portion of the load. With SNAT, servers return traffic to the source SE MAC. SEs forward response traffic directly back to clients. SE 1
  • 23. Copyright Avi Networks 2018 Scale Service Engines via Upstream Router • All SEs advertise the VIP to BGP via Route Health Injection • Router hashes client flows across SEs • ECMP mode enables scaling across 2 to 64 Service Engines • With SNAT, servers return traffic to the source SE MAC address • SEs send response traffic directly to clients Failure Mitigation • BFD may be enabled to ensure faster detection of an SE failure • Persistence and SSL connections are mirrored to ensure a graceful and automatic recovery in case of a router hash redistribution • SEs will forward incorrectly hashed flows to the proper SE SE ECMP Scaling SE
  • 24. Copyright Avi Networks 2018 SE Auto Scaling SE 1 Scaling • Scale Out • Scale In – Gracefully remove an SE from the active/active group – Waits one minute for connections to close before scaling in • Migrate 1. Scale out from SE1 to SE2 2. SE2 GARPs for the VIP 3. Scale in to SE2, removing SE1 from servicing the VIP Manual Scaling • Administrator initiated scale in, out, and migrate • Default mode Auto Scaling • SE Group may be configured for manual or automatic scaling • Avi does not [yet] recommend auto scaling – Works for CPU above/below threshold – Auto scale available via CLI/API
  • 25. Copyright Avi Networks 2018 Scale SE Performance Up and Out SE SE SE SE SE SE Scale up with more CPU cores Scale out with more SEs
  • 26. Copyright Avi Networks 2018 Multi Availability Zones for Public Cloud • Public clouds such as AWS split a data center into three Availability Zones • Each AZ is a separate IP network space • AWS customers are expected to load balancing traffic into the three Azs • Avi deploys an SE per AZ • DNS is then used to distribute traffic across the three VIP addresses for an app • The Avi Controller removes a VIP from DNS if that AZ or SE is down • Multi AZ awareness for AWS and Azure require a DNS profile for the cloud AZ 1 AZ 2 AZ 3 www.avi.com 20.1.1.1 20.2.2.2 20.3.3.3 Traffic distribution in AWS data center with 3 AZs
  • 27. Copyright Avi Networks 2018 Bare Metal Virtualized Containers Public Cloud CONTROLDATA MESOS Hybrid Cloud Both traditional and modern use cases Automation Highly programmable, plug-n-play Analytics Actionable insights key to automation Separate Control and Data Plane Manage as one, not many devices • Why change the HA model? • Active/Standby is based on a physical, device-centric world • Doesn’t scale, increases management complexity • How has the model changed? • NFV model, Active/Active • Specific examples of impact • Nearly infinite scale • Easier management, easier to write automation
  • 28. Copyright Avi Networks 2018 Next Steps • Avi Tech Corner Webinars avinetworks.com/webinars-avi-tech-corner • Avi Knowledge Base avinetworks.com/docs • Avi Workshops avinetworks.com/workshops • Virtual Lab email: education@avinetworks.com
  • 29. Copyright Avi Networks 2018 Nathan McMahon education@avinetworks.com avinetworks.com/workshops

Editor's Notes

  1. Which mode is the illustration referencing? (N+M) In N+M mode, what are the values illustrated for N & M?
  2. Having 1 very large SE means we need a second SE for HA. So it starts to look like an A/S HA model. Having lots of smaller SEs means minimal overprovisioning to account for HA. Typically customers should fall somewhere in the middle ground of these two extremes
  3. Multi AZ is a critical requirement for public clouds. Students need to make their site be AZ aware. Instructor needs to add a DNS VS, set it as the Avi Administration > Settings > DNS Service: “DNS VS” Then add AZ2 and AZ3 IPs to the VS. If the VS were created before DNS VS existed, they will get an error message about needing to add the FQDN name to the VS.