Basic Planning
Learn about the best practices to get started with managing infrastructure for Oracle Communications Unified Assurance.
Define Business Requirements
To ensure Unified Assurance provides the value for which it has been purchased, it is extremely important to define the business goals extensively and communicate and measure against them throughout the duration of the implementation. Examples of business goals are:
-
Automation for operations
-
Automation for administration
-
Increased visibility using dashboards and diagrams
-
Replacement of previous tool sets
-
Cost reduction and avoidance, return on investment
Operations
Operations, in the context of Unified Assurance, are the usage or interaction of the various user communities with the Unified Assurance software solution. This may include executives, engineers, end-users, and NOC operators.
User Authentication
Unified Assurance supports the following types of user authentication:
-
Internal authentication
-
External authentication with:
-
Microsoft Active Directory: Primary and secondary hosts, domain suffixes, and secure connections.
-
Open LDAP: Primary and secondary hosts, distinguished names, and secure connections.
-
SAML
-
You can configure multiple instances of each of the external authentication types, and external users map to internal Unified Assurance users. Transient users (users authenticated externally that do not have corresponding internal Unified Assurance users) are also supported.
See the following in Unified Assurance Security Guide for more information:
Multitenancy
Unified Assurance supports the capability to provide multiple securely segmented read-only access groups used for multitenancy. This is usually for client-side portaling or segmented departments within a single organization. You define these configurations by using Unified Assurance user groups, device groups, dashboard groups, diagram groups, and event filter groups. You must define multitenancy configurations before implementation to prevent rework.
For more information, see:
-
Understanding Unified Assurance Multitenancy in Unified Assurance Concepts.
-
Configuring Multitenancy in Unified Assurance Security Guide.
-
Authentication Options and Adding User Accounts in Unified Assurance Security Guide.
Defined Roles
To leverage any secure application, you must define roles and assign privileges to them. Unified Assurance includes five default roles:
-
Administrator
-
Anonymous
-
API
-
Operator
-
Publisher
Determine if you need any additional roles at the earliest stage of implementation.
User Group Creation
Like roles, you must determine user groups before implementation. User groups use roles to control access into Unified Assurance, and have associated groups of users.
Process Management
Oracle recommends that you list and create use cases for all major processes (defined or yet undefined) if interaction with the Unified Assurance software solution is required or wanted. This ensures that any process interaction will be successful and accomplished with ease.
Unified Assurance Architecture
Unified Assurance architecture consists of a typical three-tier implementation, where the presentation layer leverages an Apache web server, the data layer consists of MySQL, OpenSearch, Influx, and Neo4j database instances, and the application layer runs Unified Assurance compiled binaries.
Components Required
Before determining your architecture, you must first determine all the Unified Assurance components you will need to achieve the features and functionality you require. This list is vital in the planning and design phases and any architecture created without this component list completely identified may run into scaling issues.
Security Requirements
Determine if you require a FIPS 140-2 compliant environment. Unified Assurance installed on Oracle Linux 8 supports FIPS 140-2 compliance.
See FIPS Compliance in Unified Assurance in Unified Assurance Security Guide and FIPS 140-2 Compliance in Oracle Linux 8 in Oracle Linux 8 Enhancing System Security for more information.
System Requirements
Oracle recommends the following:
-
A dedicated partition for the Unified Assurance installation. By default, this is the /opt partition.
-
RAID 10 for any database or bulk loading uses.
-
Configuring and synchronizing Network Time Protocol (NTP) across all servers.
-
Monitoring all Unified Assurance servers for CPU, Disk, Memory, and IO at least every five minutes.
Hardware Requirements
Estimates are broken out into the following categories, with baseline hardware recommendations for Unified Assurance:
-
Minimum: Less than 1000 devices. See Minimum Installations.
-
Medium: Between 1,000 and 25,000 devices. See Medium Installations.
-
Large: Greater than 25,000 devices. See Large Installations.
-
Extra large: For extremely large installations and more specific requirements, contact your Oracle Sales and Consulting representative.
All Historical database storage estimates are based on default configuration and retention policies. Storage requirements depend on data retention policies, the number of raw events received, and the number of metrics collected. Contact your sales representative for a personalized estimate.
Minimum Installations
Minimum installations are less than 1000 devices.
Supported implementations:
-
One server: All-in-one installation
-
Two servers: A presentation and database server, and a separate collection server (including a microservice cluster)
-
Four or more servers: Separate presentation server, database server, Historical database server, and collection server (including a microservice cluster)
Requires:
-
Linux server support
-
Intel/AMD x86_64
-
Oracle Linux or Enterprise Linux 8.x (64-bit x86) ("Base" OS install)
-
4 core (including hyper-threaded or vCPU) Xeon E5 2GHz or better processors
-
32 GB RAM
-
500 GB drive space for application and logs
-
2 TB historical storage (database server only)
Medium Installations
Medium installations are between 1,000 and 25,000 devices.
Supported implementations:
-
Six or more servers: Separate presentation server, database server (Event, Metric, and Graph databases), Historical database server, and three collection servers (including microservice clusters)
-
Redundancy doubles the number of servers
Requires:
-
Linux presentation/collection server support
-
Intel/AMD x86_64
-
Oracle Linux or Enterprise Linux 8.x (64-bit x86) ("Base" OS install)
-
8 core (including hyper-threaded or vCPU) Xeon E5 2GHz or better processors
-
32 GB RAM for each server
-
500 GB drive space for application and logs
-
2 TB historical storage (database server only)
Large Installations
Large installations are greater than 25,000 devices.
Supported implementations:
-
Eight or more servers: Separate presentation server, Event database server, Metric database server, Graph database server, Historical database server, and three collection servers (including microservice clusters)
-
Redundancy doubles the number of servers
Requires:
-
Linux presentation/collection server support
-
Intel/AMD x86_64
-
Oracle Linux or Enterprise Linux 8.x (64-bit x86) ("Base" OS Install)
-
16 core (including hyper-threaded or vCPU) Xeon E5 2GHz or better processors
-
64 GB RAM for each server and 128 GB RAM for database servers
-
500 GB drive space for application and logs
-
2 TB historical storage (database server only)
Software Map
To determine the software map, apply the components list to the hardware list. Although you can install components on any server, Oracle recommends the following best practices:
-
Pollers of any type should not be on the same server where the web server resides. This may cause user interface interference.
-
Components that poll an internal historical storage database (threshold engine, historical event rotation, and so on), should reside on the same server as the Historical database.
-
Metric pollers and threshold engines are highly memory dependent.
-
Metric pollers are scaled by average response time to device polled, percent of devices down, timeout settings, threads defined, historical database insert speed, CPU speed and availability, and memory availability.
-
Because database instances (Metric, Event, Historical, and Graph) may have high disk requirements, you can install them on separate servers.
-
Install metric pollers and discovery agents on the same server. This prevents discovery and polling mismatch failures.
Redundancy, Failover and Failback
Unified Assurance supports pairs of geographical or cross-data center (XDCR) failover and failback capabilities. Oracle recommends you read the redundancy brief provided by Oracle and document your failover requirements for production implementation before implementation.
See Scalability and Redundancy in Unified Assurance Concepts.
Backup and Restore Strategy
As with all software, standard backup and restore requirements should apply. By default, the Unified Assurance configuration database is backed up every Saturday at 1 AM. You must create a backup plan for the other databases and define all backup requirements before implementation.
See Backup and Restore in Unified Assurance System Administrator's Guide for more information.
Unified Assurance Component Network Requirements
Unified Assurance uses some network ports for communication between components. You must open them bi-bidirectionally in your local operating system firewall and network ACLs or firewalls. Oracle recommends doing this before installing Unified Assurance.
-
Unified Assurance Message Bus TCP/5671 (Presentation)
-
Unified Assurance MySQL Database TCP/3306 (Presentation/Event Database)
-
Unified Assurance Web Server TCP/80,443 (Presentation)
-
Unified Assurance File Synchronization TCP/8873 (Presentation)
-
Unified Assurance Redundancy Wizard TCP/8055,8056 (Database)
See Linux Prerequisites for a detailed list of ports that are used.
Unified Assurance Typical Collection Network Requirements
Because Unified Assurance is typically used for polling devices across the network, you must open the following ports from at least the collection servers to the remote network devices. Oracle recommends creating a management subnet to make maintenance easier.
-
ICMP ping or Generic Up/Down polling requires ICMP connectivity from servers to devices
-
SNMP polling requires network connectivity (UDP/161) from servers to devices
-
WMI/NT Event log Polling requires network connectivity (TCP/135) from servers to devices
-
Management Servers and monitored devices need to contact the NTP server (UDP/123)
-
Syslog collection requires network connectivity (UDP/514) from devices to servers
-
SNMP Trap requires network connectivity (UDP/162) from devices to servers
-
MTU Packet Size should allow for 1475 or higher
Scale and Capacity Planning
Unified Assurance scales based on two major areas: the number of database inserts per second and database retention requirements. These can be calculated using the Unified Assurance scalability brief and storage calculation spreadsheet. A detailed discovery and planning session is required to determine your polling requirements and storage retention levels for the initial implementation as well as considering future capacity required or expected.
DNS Caching
DNS caching on Linux is disabled by default, which can cause performance issues (such as increased load on DNS servers and increased response latency) in large scale implementations. For optimal performance, Oracle recommends configuring a DNS caching agent on all Unified Assurance servers.
Recommended DNS Caching agents:
Device Management Process
The device management process is one of the most common key administration processes that will need to be defined. This process includes what needs to be done in Unified Assurance to facilitate adding new devices to be monitored as well as how devices are decommissioned and handling the various changes a device may experience. Below is a basic list of areas that need to be addressed by the process:
-
Auto Discovery: Devices being entered into Unified Assurance's Core Device Catalog
-
Auto discovery Scheduled Job (Ping Scans, Seed Lists, CDP, LDAP)
-
Collection Discovery by collector rules
-
Manual Entry in the Unified Assurance UI
-
-
Device Typing: Setup of Device communication and device type/category (SNMP or WMI)
-
SNMP Discovery agent: v1, v2, v3 community strings (v1 polling not generally recommended)
-
WMI Discovery agent: Windows only
-
-
Instance Discovery: Discovery of sub-devices (Interfaces, IP SLA probe tags, Disk Partitions)
-
Automated by Metric Manager Discovery agents
-
Manually in the Unified Assurance UI
-
-
Metric Discovery: Setup of polling metric instances
-
Automated by Polling Policy Discovery agent
-
Manually in the Unified Assurance UI
-
-
Network Inventory Collection: Interface, ARP, IPRoutes, CDP, and so on.
-
Topology Stitching: Standard Agents for IPRoutes, ARP, and CDP, otherwise customization is needed
As part of this process there may be additional enrichment or security required rules that may need to be written. The following are commonly required:
-
Device Group placements (auto.rules, snmp.rules)
-
MetaTags created (auto.rules, snmp.rules)
Most of the discovery engines are designed for discovery of new devices. The following device changes may need to be handled in some way by this process:
-
Device DNS Name changes (Auto discovery agent automatically updates)
-
Device IP Address changes (Manual or customization)
-
Device Security changes (Community Strings) - SNMPDiscovery or manual
-
Device Type/Category changes - Manual or customization
-
Device Group changes - customization
-
Device Polling requirements - Automated with pollers, customization for collectors
-
Sub-device changes (New, changes, decommissions, etc)
-
IP SLA Probe Tags - Manual or customization
-
Interfaces - Manual or customization
-
Partitions - Manual or customization
-
Topology - Manual or customization
-
Inventory - Automated with -Flush option
-
When devices are decommissioned the following areas will need to be addressed:
-
Device Catalog Deletion - Bulk Load or via SQL
-
Retention of Metric Data - Static Devices
-
Retention of Topology
It is also important to understand how discovery and/or re-discovery processes will be initiated. Typically, most customers run discovery on-demand, but you may wish it to be fully automated and scheduled jobs can be enabled to run at set time of day, week, and month.
Dashboards
Custom Dashboards
-
List of Dashboards that need to be created with the following information
-
Audience (who is it designed for)
- Customers, Executives, NOC Operators
-
Security restriction (who can access it, etc.)
- Customers, Executives, NOC Operators
-
Contents
- Background, Metrics, Filters, Images, etc
-
Layout requirements (where does what go?)
-
Maintenance requirements (who is going to keep it up to date)
- Admin, NOC Operators
-
Plug-ins
- Who should have access to the Topology/SLM Plug ins?
Metrics (Polling, Collection, Display)
Metric Manager is used for collection, processing, reporting, and displaying of availability, performance, or capacity-based data. The individual component types need to be properly scoped so that the correct configuration can be deployed. Below are commonly discussed requirements that need to be discussed/determined before moving into implementation.
Base Configuration
-
TopN Overviews - Type/Scope/Direction
-
Custom Collection Definitions (Name, How, Maximums, Factor, Scales/Abbr)
-
Retention (Raw, Hourly, Daily) - reference scale brief/calculator
-
Raw - default 14 days
-
Hourly (1hr consolidation) - default 3 months
-
Daily (24hr consolidation) - default 2 years
-
Pollers
-
What polling intervals are required?
-
Every minute?
-
Every 5 minutes?
-
-
Clustering/Location
-
How many devices need to be polled?
-
What is the average latency/packet loss for the devices to be polled by location?
-
Is load balancing required? How many poller instances are required?
-
-
Database Scalability
-
What are the expected inserts/sec for each poller and each database instance?
-
What polling delay is expected/required?
-
Which storage engines will be used by pollers? Direct or Bulk Insert?
-
-
Custom SNMP Polling
- What MIBs and Objects are required for polling (Gauges, Indexed tables, and Counter)?
Collectors
-
What custom collection feeds are required?
-
Format for rules parsing (Are formulas required?)
-
Feed Type/Protocol
-
-
How many interfaces, of what capacity, need to be collection targets?
Synthetic Transactions
-
How many types of transactions need to be run? How many concurrently? How many POVs?
-
Are the transactions run via command-line or WinTask policy?
-
How many instances of the same transaction type need to be run?
-
Do the transactions need to be setup to run automatically?
-
Do you have code examples and test-beds available and documented?
Thresholding
-
Do you need threshold notifications or just event integration?
-
Approximately how many thresholds will need to be defined?
-
Do you have defined service levels that threshold differently?
-
Do you need to threshold metric trends across daily, weekly, monthly, or yearly change?
-
Do you need dynamic thresholding for metrics via cyclical deviation?
-
What are your static threshold levels that need to be defined, i.e.
-
CPU: > 75% Warning, > 90% Critical
-
Bandwidth: > 75% Warning, > 90% Critical
-
Dynamic Device Overviews
Besides the following, are there any custom metric overviews required?
-
Device Health (Ping Availability, CPU, Disk, Memory)
-
Network Traffic (Latency, Packet Loss, Interface Stats)
-
IPSLA by Probe Tag/Type (Availability, Latency, Jitter, Packet Loss)
-
Response Times (Synthetic Transaction Latency)
Custom Reporting
-
What scheduled static reporting templates are required?
-
Are they Graphical Reports?
-
Trend Report
-
Time-Series
-
-
Are they List Reports?
-
What schedules or distribution lists are assigned to each graph template?
Service Level Management
-
What services need to be monitored? Quantity?
-
VOIP
-
DNS
-
WEB
-
-
Are they documented?
-
Who will have access to the SLM data?
-
How often are they changed?
-
Is automated creation/maintenance required?
Events
Event Collection
-
What are the event feeds that need to be collected?
-
Format
-
Feed/Protocol/Method
-
-
Which aggregators will be needed?
-
Standard Trap/Syslog (forwarding?)
-
API-based: NNM, etc
-
TL1: Gateway/NE list - Chat In/Out Login, Poll, Logout scripts
-
Generic (Database, Socket, Flat File, Command, Email, etc.)
-
-
What are the different collection locations and how are they split up?
Scalability
-
What is the estimated events/sec per aggregator?
-
How many total events/sec are required for the historical event repository?
-
How many days of historical event data do you want to keep?
Escalation/Notification
-
What simple notification profiles are required? (Uni-directional without time delay)
-
Do you have operational performance monitoring requirements that you want to use iE&N (Intelligent Escalation and Notification Engine) for? (one-to-one mapping of event to responsible group)
-
What are your escalation tree requirements by type?
-
Do you want to use the rotating on-call engine?
-
Who is going to maintain the on-call list?
-
What are the Users/Groups/Shifts/Schedules?
-
-
-
What custom notification requirements are needed? (SMS, Voice, IM, other, etc.)
-
What text email formats are required?
-
What SMTP servers do you intend to use?
-
Do you intend to use the email acknowledgment engine? What POP3 server, port, user, password?
Event Display
-
What default Filters are required?
-
Device filter
-
Specific Alarm Types
-
Time frames
-
Public/Private to Groups?
-
-
What Filter Group configurations need to be created?
-
Which Filters belong in which groups?
-
Which Filters can specific groups use?
-
-
What default Displays need to be required?
-
Public/Private to Groups?
-
What Fields, Names, and Order
-
-
What custom right-click tools need to be created? Who uses them?
-
SQL queries
-
CGI scripts
-
-
Each non-read-only user group uses its own menu defined. Are there additional menus you want created? List them out fully.
Enrichment
-
What Internal Event Enrichment is required?
-
Device Catalog (Priority, Meta Tags, etc.)
-
Topology (Next Hop, etc.)
-
Inventory (Interface Name, etc.)
-
-
What External Event Enrichment is required?
-
Format/Mappings
-
Feed/Method (Flat File, Database, Socket, Command-line)
-
-
What Method for each enrichment policy makes the most sense?
-
Rules-based (quick/easy/real-time but keeping up to date may be a problem)
-
Agent/Connector (not real-time, but fewer look-ups, less maintenance required)
-
Correlation
-
What correlation requirements beyond the basic generic Clear and Expire do you require?
-
Heart beating
-
Stacking or Disparate Events: This + This = That (Same Source/Different Source)
-
XinY - Event Thresholding to new summation event?
-
Parent/Child or RCA: Topology-based? Data Source?
-
Downstream Suppression: If this is downstream from that, and that is in alarm, suppress this (What is suppressed?)
-
Custom Policy/Automatic Remediation?
-
Service Impact Analysis
-
-
Correlation Methods
-
Rules (Correlation by De-duplication, Direct Query, etc.)
-
Mechanizations (Database Stored Procedures)
-
Connector/Agents (Custom Single Policy, Time-of-Day Correlations)
-
CAPE (Multiple Policy Engine, Auto-Remediation)
-
PRCA Connector (Network-based Topology Correlation, Down-stream Suppression)
-
SLM Connector (Real-time Service Impact Analysis)
-
Northbound Integrations
(Custom, Ticketing, Historical)
-
What Ticketing Integration is required?
-
When/How do you want to ticket? (Methods)
-
How to integrate with the ticketing system? (API, Socket, Database, Flat File, Command-line, SOAP, etc.)
-
What scheduled synchronization requirements do you have?
-
Work Log + Alarm Journal
-
Ticket Status + Event Severity/Acknowledgment
-
Other?
-
-
-
What other northbound integrations are required? (Historical/Other)
Historical Reporting
-
What table-based historical event reports are required?
-
Query (Where Clause)
-
Format (Columns)
-
Schedule to run?
-
-
What other custom reporting is required?
Topology
Network Inventory Collection
-
Do you want to use CDP-based data? Is it enabled on all routers/switches/servers?
-
How often do you want to collect inventory (Discovery vs. Re-Discovery)?
-
Do you need to split up your inventory collection to collect from a certain location, time of day, etc.? Do you want to limit the collection and exclude certain data?
Topology Stitching
(ARP, CDP, IPRoutes, Config Agent, Custom CRM/Provisioning, Billing)
-
Do you wish to still keep the following default inventory data?
-
ARP
-
CDP
-
IPRoutes
-
-
What custom Config Agent stitching do you need to collect?
-
What custom 3rd party stitching data do you need to collect?
-
Format/Mapping
-
Feed/Method (Database, Flat File, Socket, Subversion, SOAP)
-
Configuration Agent
-
What Devices are targets? How are they grouped?
-
How often do you want them to be collected?
-
What differential notifications do you want? Where do you want the notifications sent?
-
What Chat In/Chat Out scripts need to be developed and tested?
-
Commands/Responses
-
User/Passwords
-
DNS IP SLA Naming Standards and Examples
A standard IP addressing scheme for distributing IP networks between access, distribution, core, WAN networks, and data centers should be devised. Networks should be allocated in CIDR blocks to help routing protocols summarize the networks as well as provide ease of identification.
All host and network devices should be configured to use Domain Name Service (DNS) servers for name resolution. For Unified Assurance components, configuring both forward and reverse resolution is recommended.
At least 2 DNS servers should be setup to provide authoritative DNS lookups to devices and hosts. Zones should be setup to provide forward lookups for all domains and subdomains (example.com), as well as provide reverse lookups for all IP networks (X.X.X.in-addr.arpa). If one device is accessible from multiple interfaces/IP addresses (from an ICMP and/or SNMP perspective), you should add reverse DNS to the management addressed host name.
A standard naming scheme should be used to provide consistency. The standard can include description by location, device type, purpose, etc. Names cannot include spaces, nor should they contain underscores (_) due to standards-based DNS restrictions.
Example 1
(Device Use)(Device Type)(Floor)(Device Number)(BuildingName)(Riser)
ASW-1-2-RDP-A1 = Access Switch 1st Floor Device Number 2 Galter Riser A1
DSW-1-2-RDP-PACS = Distribution Switch 1st Floor Device Number 2 Feinberg Building Riser PACS
CSW-5-1-259 = Core Switch 5th Floor Device Number 1 259 Building Riser N/A
Example 2
(Building)(Floor/Closet)(Device Type)(Device Number)
1st 3 characters designate building
4th thru the 7th characters designate the area, floor and closet
8th thru the 10th characters designate the device use & function
11th and 12th characters designate the device number
The example would result in a device name like "str-x201-wap-01", and could translate to meaning "the 1st wireless access point at stratford, area x, 2nd floor, 1st closet".
Cisco
Example IOS Configuration
ip domain-name <DNSDomain>
ip name-server <DNSSrv1>
ip name-server <DNSSrv2>
Example CatOS Configuration
set ip dns domain <DNSDomain>
set ip dns server <DNSSrv1> primary
set ip dns server <DNSSrv2>
set ip dns enable
Basic Device Configurations
Common Server Configurations
The following are some of the common server configurations for Unified Assurance.
SNMP Polling Options
Oracle recommends leveraging the Net-SNMP agent for Unix, Linux, and Windows-based platforms. For more information about the Net-SNMP agent, see the Net-SNMP website. If you wish to use Windows SNMP Agent, we highly recommend downloading SNMP Informant Patch for Windows. The stability, popularity, conformity of the agent, various platforms supported, and large default MIB support are the reasons for this recommendation. Oracle provides a net-snmp SNMP rules file that provides many of the common metric data requirements customers are seeking. For all other agents, custom SNMP rules may need to be created, so you may want to consult with your Oracle sales representative before implementation. Be aware that only SNMP v2 and v3 are supported by default by the pollers.
Unix/Linux Syslog Forwarding
Syslog forwarding is documented by the syslog daemon being used on your server. Some external research about your particular operating system will be needed, and you can also run "man syslogd.conf".
NTP
Oracle recommends all devices being monitored are synchronized to a single common clock source. Consult with your operating system documentation to determine the best NTP strategy for your organization.
You can use the chrony suite to configure network time. See Configuring Network Time in Oracle Linux 8 Setting Up Networking for more information.
Common Network Device Configurations
The following are some of the common network device configurations for Unified Assurance.
SNMP Polling Setup
Most network devices monitored by Unified Assurance leverage SNMP for communication. Consult your vendor documentation on how to set up SNMP connectivity and test the process before implementation. Below is a running list of different vendors and the recommended configuration required:
Cisco
Example IOS Configuration
no snmp-server community public RO
no snmp-server community private RW
snmp-server community <enter_read_only_password_here> RO
snmp-server community <enter_read_write_password_here> RW
Example CatOS Configuration
set snmp community read-only <enter_read_only_password_here>
set snmp community read-write <enter_read_write_password_here>
set snmp community read-write-all
Cisco devices can provide serial numbers through SNMP, although they are only configured by default on fixed chassis devices. All other modular devices should have the serial number configured manually to provide ease of contract and warranty management.
Example IOS Configuration
snmp-server chassis-id <enter_device_serial_number_here>
Example CatOS Configuration
set logging level snmp 7 default
Interface Descriptions
To assist with troubleshooting issues and to provide additional information to an enterprise management system, all interfaces should have a description defined for them. This description can either be a generic description about the interface type, or can include specific information like customer name, circuit ID, etc.
Example IOS Configuration
interface Serial0/1
description T1 connection to Internet provider (circuit id: XXXX/0000001)
Example CatOS Configuration
set port name 4/1 Trunk to distribution switch
SNMP Interface Index Renumbering
One of the big challenges associated with many SNMP performance management tools is the tool's ability to handle interface index re-numbering. It is true that some tools handle this better than others; however, there are steps that your organization can take to minimize the problems you'll likely encounter when this occurs. One of the relatively easy and straight forward approaches to take is to configure all IOS routers with the command 'snmp-server ifindex persist'.
The tools also need to be smart enough to know when ifIndexes change for other devices. Usually vendors say they key off of the ifDescr, but they should allow this to be configurable by device type. For example, ifDescr is always the same on CatOS Ethernet ports, but portName will be unique.
Syslog/Trap Forwarding
Most network devices monitored by Unified Assurance forward event information either by Syslog or by Trap. Some devices can send equal quality of data in either format, and where this is the case Oracle recommends using syslog because it is easier to parse. Consult your vendor documentation on how to set up syslog or trap forwarding and test the process before implementation. Below is a running list of different vendors and the recommended configuration required:
Cisco Syslog
Example IOS Configuration
logging trap notifications
logging source-interface <DevMgmtIntf>
logging <MgmtSrvIPAddr>
Example CatOS Configuration
set logging server severity 5
set logging server enable
set logging server <MgmtSrvIPAddr>
set logging timestamp enable
NTP
Oracle recommends all devices on monitoring be synchronized with a single common clock source. Consult with your vendor documentation to determine the best NTP strategy for your organization. Below is a running list of different vendors and the recommended configuration required:
Cisco
Example IOS Configuration
service timestamps debug datetime msec localtime show-timezone
service timestamps log datetime msec localtime show-timezone
clock timezone CST -6
clock summer-time CDT recurring
ntp authenticate
ntp authentication-key 1 md5 <enter_auth_key_here>
ntp trusted-key 1
ntp source <DevMgmtIntf>
ntp server <NTPSrv1> key 1
ntp server <NTPSrv2> key 1
Example CatOS Configuration
set timezone CST -6
set summertime enable CDT
set summertime recurring
set ntp broadcastclient disable
set ntp client enable
set ntp authentication enable
set ntp key <enter_auth_key_here>
set ntp server <NTPSrv1>
set ntp server <NTPSrv2>