COMP13: The WeakestLink in your High Availability system

1
COMP-13 The Weakest-Link in your High
Availability system

• How to make sure your HA is really highly
  available

Hugo Loera
Senior Technical Support Engineer
2
Goals

• To consider the entire system architecture for
  High Availability (HA).
• Provide a summary of all the fault tolerance
  features that provide HA.
• To think of possible disaster scenarios.

3
Agenda

• Your system chain
• Keeping your hardware protected
• Keeping your network connected
• DB availability
• Protecting WebSpeed and AppServer tiers
• Sonic fault tolerance features

4
Your system chain

• The strength of a computer system is based on
  its weakest link, much like a chain.
• Goal to remove all of the single points of
  failure for an application.

Facilities
Hardware
Software
Network
User
5
Your system chain
6
Your system chain

• It is necessary to provide redundancy

Facilities
Hardware
Network
SonicMQ
Internet
WebSpeed
Sonic Adapters
Application Servers
Data Base
7
Agenda

• Your system chain
• Keeping your hardware protected
• Keeping your network connected
• DB availability
• Protecting WebSpeed and AppServer tiers
• Sonic fault tolerance features

8
Keeping your hardware protected
9
Keeping your hardware protected

• Hardware
• Storage Devices
• RAID, Remote data replication device
• CPU
• Cluster hardware
• Facilities
• Air conditioning
• Fire and flood protection
• Power Source
• UPS, Power Plant, Alternative Power Supplies
• Alternate site

10
Keeping your hardware protected

• RAID (redundant array of independent disks)
• Striping (provides load balancing)
• Mirrored configurations (for increased
  reliability)

RAID Configuration
Fiber Channel
11
Keeping your hardware protected

• HA Cluster

Node 2
Node 1
Application
Application
OS
OS
Cluster Manager
Cluster Manager
Cluster name or Cluster IP Address
Local Storage
Local Storage
OE Enterprise
OE Enterprise
12
Keeping your hardware protected

• Cluster

Node 2
Node 1
Application
Application
OS
OS
Cluster Manager
Cluster Manager
Cluster name or Cluster IP Address
Local Storage
Local Storage
OE Enterprise
OE Enterprise
13
Agenda

• Your system chain
• Keeping your hardware protected
• Keeping your network connected
• DB availability
• Protecting WebSpeed and AppServer tiers
• Sonic fault tolerance features

14
Keeping your network connected
Clustered Servers
15
Keeping your network connected

• Network
• Alternate network paths
• Redundant network links
• Configure a Domain Name Server (DNS)

Site A
Site B
Users site
16
Keeping your network connected

• If using alternate site
• Use a virtual DNS entry for
• Your DB server
• Application server
• Web servers
• Programs File Server
• Hostnames should be virtual names
• Consider user VPN access for emergencies

17
Agenda

• Your system chain
• Keeping your hardware protected
• Keeping your network connected
• DB availability
• Protecting WebSpeed and AppServer tiers
• Sonic fault tolerance features

18
DB availability
Clustered Servers
Multiple links
19
DB availability

• Cluster configuration
• Protects from hardware failure
• Quick recovery time
• Easy to register DB once the cluster has been
  configured.
• OpenEdge cluster support
• Included with the Enterprise RDBMS
• Allows use of the procluster command to
• register the DB with the cluster manager
• disable DB registration
• start and stop the DB
• query the status of the DB

20
DB availability

• Vendor specific cluster commands

21
DB availability
Caa_stop
..stop
Node 1
Cluster Manager
procluster DB stop
pscluster
OE Enterprise

• OE provides a common interface
• For different cluster products.
• To integrate OE commands into the cluster
  environment.

22
DB availability

• OpenEdge DB replication
• Protects from storage problems
• Site fault tolerance protection
• Can be manually activated

Client
Client
Target
Source(live)
23
DB availability
Combining OpenEdge Replication with Clusters
Replication Agent
Replication Server
Node 1
Node 2
Target DB
OpenEdge Enterprise Replication
DB Broker
Operating System
Operating System
Source DB
Cluster Manager
Cluster Manager
Common Storage
Site B
Local Storage
Local Storage
OpenEdge Enterprise DB Replication
OpenEdge Enterprise DB Replication
Site A
24
DB availability
Combining OpenEdge Replication with Clusters
Replication Agent
Replication Server
Node 1
Node 2
X
Target DB
OpenEdge Enterprise Replication
DB Broker
Operating System
Operating System
Source DB
Cluster Manager
Cluster Manager
Common Storage
Site B
Local Storage
Local Storage
OpenEdge Enterprise DB Replication
OpenEdge Enterprise DB Replication
Site A
25
DB availability
Combining OpenEdge Replication with Clusters
DB Broker
Node 2
Node 1
X
X
X
DB
OpenEdge Enterprise Replication
Operating System
Operating System
Source DB
Cluster Manager
Cluster Manager
Common Storage
Site B
Local Storage
Local Storage
OpenEdge Enterprise DB Replication
OpenEdge Enterprise DB Replication
X
Site A
26
Agenda

• Your system chain
• Keeping your hardware protected
• Keeping your network connected
• DB availability
• Protecting WebSpeed and AppServer tiers
• Sonic fault tolerance features

27
Protecting WebSpeed and AppServer tiers
Clustered Servers
Multiple links
Cluster Replication
28
Protecting WebSpeed and AppServer tiers

• NameServer Fault Tolerance
• Using UDP broadcast address
• Will provide instant access to alternate brokers
• Can use Name Server Load Balancing
• Redundant brokers can be registered
• AppServer brokers
• WebSpeed brokers

29
Protecting WebSpeed and AppServer tiers

• Name Server Replication based fault tolerance

Client
LAN 1 Gb Backbone
Host2
Host1
NS1
NS1
Subnet
AppServer Broker
AppServer Broker
WebSpeedBroker
WebSpeedBroker
30
Protecting WebSpeed and AppServer tiers

• Getting the UDP broadcast address on UNIX

netstat -i Name Mtu Net/Dest Address
Ipkts Ierrs Opkts Oerrs Collis Queue lo0
8232 loopback localhost 2318894 0 42318894 0
0 0 le0 1500 psc-bed oasis
3625559 16 79807506 0 0 0
ifconfig le0 ce0 flags1000843ltUP,BROADCAST,RUN
NING,MULTICAST,IPv4gt mtu 1500 index 2 inet
192.128.5.30 netmask ffff0000 broadcast
192.128.255.255
31
Protecting WebSpeed and AppServer tiers

• Getting the UDP broadcast address on Windows

C\gtipconfig Windows IP Configuration Ethernet
adapter Local Area Connection        
Connection-specific DNS Suffix  .
blabla.speak.com        IP Address. . . . . . .
. . . . . 192.128.3.24        Subnet Mask . .
. . . . . . . . . 255.255.0.0        Default
Gateway . . . . . . . . . 192.128.3.1
IP address 192.128. 3. 24
Subnet Mask 255.255. 0. 0

__________________ UDP Broadcast
IP address 192.128. 3. 24 NOT
Subnet Mask 0. 0.255.255 OR

__________________ UDP Broadcast
192.128.255.255
32
Protecting WebSpeed and AppServer tiers
NAME SERVER REPLICATION
Connect(-H 192.128.255.255 -S 5162 -AppService
asbroker1)
Client
LAN
Host2
Host1
NS1
NS1
NameServerNS1 portNumber5162 hostName192.128.25
5.255
Subnet
AppServer Broker
AppServer Broker
33
Protecting WebSpeed and AppServer tiers

• Using redundant WebSpeed brokers
• Messenger has to use UDP broadcast address
• Distributed standalone messengers can be
  configured
• Can use Name Server Load Balancing
• Can use multiple web servers
• For fault tolerance
• Load balancing
• Multiple products are currently available for web
  server fault tolerance and load balancing

34
Protecting WebSpeed and AppServer tiers

• Stand alone WebSpeed Messenger ubroker.properties
  configuration

Sample definition of a NameServer NameServ
er.NS1 locationremote hostname192.128.255.
255 environmentNS1
srvrLogFileWRKDIR/NS1.ns.log autoStart0
. . . WebSpeed Messenger
Properties WebSpeed.Messengers
certStorePathDLC/certs/ controllingNameServe
rNS1
35
Protecting WebSpeed and AppServer tiers

• Name Server Replication based fault tolerance

LAN
Host2
Host1
NS1
NS1
Subnet
WebSpeedBroker
WebSpeedBroker
36
Protecting WebSpeed and AppServer tiers

• ABL application working with SonicMQ

Unified Broker
OpenEdge ABL.w, .p
OpenEdge Database
Sonic MQ
37
Agenda

• Your system chain
• Keeping your hardware protected
• Keeping your network connected
• DB availability
• Protecting WebSpeed and AppServer tiers
• Sonic fault tolerance features

38
Sonic fault tolerance features
Clustered Servers
Multiple links
Multiple Web Servers
Multiple ABL-Adapters
Multiple WS Brokers
Name Server Fault Tolerance
Cluster Replication
39
Sonic fault tolerance features

• What is SonicMQ ?
• A Standards-Based Enterprise Messaging Server
• User benefits
• Uses standards-based infrastructure
• Enterprise-class reliability
• Configuration monitoring
• Can include message transformation and routing
  using ESB (Enterprise Service Bus).

40
Sonic Components
Sonic fault tolerance features
Domain
Container1

• Domain
• Domain Manager
• Agent Manager
• Manager Broker
• Directory Service
• Containers
• MQ Container provides runtime support for Sonic
  components
• Can include ESB Containers to hosts ESB services
• Messaging Nodes
• Message Broker
• Broker Cluster

Queue
Topic
Container2
Queue
Topic
ESB Container
41
Continuous Availability Architecture features
Sonic fault tolerance features

• SonicMQ's CAA includes
• Resilient client connections
• Replication brokers
• Fault tolerant management services and
  communications
• Use some or all of these features to provide a
  continuously available messaging infrastructure.

42
Resilient client connections
Sonic fault tolerance features
Fault Tolerant Broker
Lost connection
JMS Client
Broker preserves client context
Recovered connection
Fault tolerant client runtime preserves session
43
Replicated Brokers
Sonic fault tolerance features

• Creates a primary and backup broker
• Backup broker replicates all acceptors on primary
• Primary accepts clients and sends replications
  data to backup
• Both brokers are functionally equivalent

44
Replicated brokers
Sonic fault tolerance features
Replication connection
ping, ping, ping,
Broker DB
Broker DB
45
Broker roles failing over
Sonic fault tolerance features
Lost connection
X
JMS Client
Replicated Broker pair
X
Recovered connection
46
Continuously available management communications
Sonic fault tolerance features

• Options include
• A cluster of management brokers
• Replicated management brokers

Recommended
47
Replicated brokers
Sonic fault tolerance features
Domain
Broker Cluster
JMS Client
Interbroker Connection
ESB Container
Node
48
Continuously available management
Sonic fault tolerance features
49
Example of a clustered management topology
Sonic fault tolerance features
Shared directory store
Backup
Primary
Primary
Backup
Cluster
50
The big picture
Site B
Site A
51
In Summary

• Consider all possible components involved in the
  Application.
• Make a plan that supports your HA configuration.
• Progress products have all the required
  technology for HA.

52
For more Information, go to

• PSDN
• DB12 Achieving High Availability with Clusters
  and OpenEdge Replication http//www.psdn.com/libra
  ry/entry!default.jspa?categoryID254externalID85
  0
• OpenEdge Replication HowTo http//www.psdn.com/li
  brary/entry!default.jspa?categoryID254externalID
  841
• Messaging Fundamentals for Sonic ESB
  http//www.psdn.com/library/entry!default.jspa?cat
  egoryID91externalID2143
• Documentation
• Open Edge 10.1B Installation and Configuration
  Guide
• OpenEdge Getting Started Application and
  Integration Services
• Progress SonicMQ Deployment Guide V7.5

53
Questions?
54
Thank you foryour time
55
(No Transcript)