Manchester University Tiny Network Element Monitor(MUTiny NEM)

A Network/Systems Management Tool

Dave McClenaghan, Manchester Computing George Neisser, Manchester Computing

1. Introduction.MUTiny Overview. - Runs on commodity hardware.

- Coded in Perl, Perl/Tk.

- Easy to install, use and maintain.

- Free (unsupported) to academia.

- Subject to ongoing development.

Network Management Overview.NMS Components:

- A Manager, running NM applications.

- A set of managed nodes. (The managed domain)

- Defined management information (MIBs).

- A Network Management protocol (SNMP).

2. The MUTiny NM Model.

Platform:

PC (Running Linux).

Description:

A Network/Systems Management tool. Coded entirely in Perl and Perl/Tk.

MUTiny Applications.

- A Network Element Monitor/Manager.

- A Unix Systems Monitor/Manager.

- A MIB data collection and reporting tool.

MUTiny NM Applications.

- Graphically display the domain topology. - Monitor and report node status changes.

MUTiny NM Applications (continued).

- Display monitored node information. - Collect and report network statistics.

MUTiny Network Management.

Key areas:

- Domain Management.

- Event Management.

- Network Statistics.

MUTiny NEM front end. _menu bar

_host attributes

_domain status

_topology display

_session bar

The GUI

2.1 Domain Management.

The Managed Domain. The set of all monitored nodes.

Monitored Node. Any network device, router, switch, etc. That is regularly polled for management information.

Domain Topology Display.

ICMP Status

(Background)

Green OK. Red No Echo. Amber Problem. Clear Unknown.

Domain Topology Display.

SNMP Status

(Foreground)

Black OK. Blue No SNMP. Grey Unknown.

Topology Management.

The initial domain topology display

Topology Management.

Add Node Window - No Auto-Discovery, nodes added by choice.

Topology Management.

- Node and ‘path’ nodes are added to the display.

Topology Management.

- Path is determined by traceroute.

Topology Management.

Change Symbol/Label Window

Connectivity Status Polling.

ICMP connectivity determined by ping. SNMP connectivity determined by sysUpTime.

Management Station

Monitored Node IF’s

ICMP

SNMP Agent

MUTiny ICMPNEM nemPoll SNMP

sysUpTime

nemStatusTable

nemPoll.conf

Configuring Status Poll Parameters.

Figure 2.2a Interface Polling Parameters

Configuring Status Poll Parameters.

Figure 2.2b General Polling Parameters - Fully configurable polling. - Poll ‘Back Off’ options.

4.2 Event Management.Event Definition.

2.2 Event Management

Connectivity Events.

ICMP:Node_No_Echo

ICMP:Node_OK

ICMP:Node_Problem

ICMP:Status_Unknown

SNMP:No_Response

SNMP:OK

SNMP:Status_Unknown

Network Event Logging.

Figure 2.3a Network Event Log Window

Network Event Alarms.

Figure 2.3b A Pop-Up Alarm

Pop-Ups may be accompanied with an optional Audible_Alarm (bell).

Event/Action Management.*

Event Configuration Window

Event/Action Management.*

Event/Source Configuration Window

Event Time Frames.

Contact/Frame Configuration Window

2.3 Network Statistics.MIB Data Collection.

Figure 2.4 MIB Data Collection Mechanism Data stored in: */nemdata/mibdata/<yearmon>/<Datafile>

Management Station

Monitored Node (hostname)

SNMP Agent

MUTiny NEM

nemDataPoll

Datafiles

nemDataPoll.conf

MIB Data

Data Storage.

Time Object Value PI 953036400 ifOutOctets.6 84118 60 953036400 ifOutOctets.7 13275 60 953037000 ifInOctets.8 7219 60 953037000 ifInOctets.5 14303 60 953037000 ifInOctets.6 18287 60

Sample data stored in: */nemdata/mibdata/200002/gw-site

Configuring MIB Data Collection.

Figure 2.5a Collection Configuration Window

Node Data Collection.

Figure 2.5b Node Collection Window

Node Data Collection. The Storage Interval. A multiple of the sampling-interval, eg.

samp-int = 60 seconds store-int = 15 min

- This greatly reduces the amount of disk space required to store the data ( by a factor of 15 in this case ).

Node Data Collection. The Store-Identifier.

The store-id, if set, stores object-id as:

<mib-obj>.<store-id>

eg.ifInOctets.Liv3

- Useful if instance-id prone to change.

Node Data Collection.

Test Collection Window

‘Change Control’.*

‘Anchors’ collection to IP_addr or Phys_addr

Reporting Network Statistics.

Figure 4.11 The Reporting Mechanism.

The reports are generated from user-defined Report Parameter Files (RPF’s).

# nemReport -r my.rpf

nemDatafiles Reports ( *.rpt )

nemReport

Report Parameter Files

Reporting Network Statistics.

The Report Template.

Reporting Network Statistics.

Figure 4.12b Sample ‘Fixed Column’ Report

- Variable and graphical* formats supported.

Report Name: liv_traffic_251099.rptReport Date: Wed Nov 10 13:00:55 GMT 1999Host Name: gw-liv.netnw.net.ukMIB Objects: ifInOctets ifOutOctets

-------InBytes.138-------- -------OutBytes.138------- -------(Bytes/sec)-------- -------(Bytes/sec)--------Date From To Average Max MaxT Average Max MaxT10/25/99 08:00 09:00 37575.64 358257.00 08:15 7254.06 31816.40 08:3010/25/99 09:00 10:00 48182.86 102154.00 09:02 11811.73 59364.60 09:3910/25/99 10:00 11:00 63443.69 123937.00 10:22 16965.79 41818.30 10:4810/25/99 11:00 12:00 109494.27 204122.00 11:38 22146.68 95139.40 11:2110/25/99 12:00 13:00 120689.33 219084.00 12:59 36755.46 77071.20 12:2710/25/99 13:00 14:00 104552.70 262792.00 13:00 39029.91 96999.50 13:2110/25/99 14:00 15:00 102345.99 187674.00 14:47 20576.39 56095.10 14:2110/25/99 15:00 16:00 93852.23 278003.00 15:19 22362.19 69550.20 15:5410/25/99 16:00 17:00 115890.41 485605.00 16:36 23883.81 57638.10 16:5010/25/99 17:00 18:00 61375.79 143225.00 17:39 16664.71 41997.50 17:31

Absolute Total: 2943.66 746.56 MBytes MBytes

4.4 Monitoring MUTiny.

Figure 4.14 The Host System Attributes Area

nemNEMPoll self checks: - NEM Processes. - Host system metrics.

MUTiny Self Monitoring.

NEM Self Monitor Configuration Window

3. WWW Cache Status Monitoring.3.1 Caching Service Configuration.

requests

responses

30 machines Manchester

3 machines L’borough

6 machines London

Global Internet

150+ Sites

Figure 3.1 The Operation of the UK National

JANET Caching Service

3.2 Caching Service Operation.

Need to know for each node:

- Network accessibility.

- CPU loading.

- Memory utilisation.

- Disk utilisation.

- Squid application status.

Network Accessibility.

Figure 3.2 Manchester Main Window

System Metric Monitoring.

Figure 3.3 Manchester Cache Systems Window

System Metric Monitoring.

Display indicates: - If the machine is pingable. - If SNMP is operational. - The CPU loading. - Memory utilisation. - Disk utilisation. - Critical process status ( squid ).

System Metric Monitoring.Prerequisites.

The UCD-SNMP mechanism

The monitored host must be running the UCD-SNMP agent software.

Management Station Remote Host

UCD-MIBsnmpd.conf

UCD-SNMP requests snmpwalk, snmpget

UCD-SNMP Agent

Access control

Host information Load Memory Disk Processes

Extensiblesection

System Metric Monitoring.

System Polling Configuration Window

System Metric Monitoring.

Disk Statistics Window

System Metric Monitoring.

A Pop_Up Alarm Optional Audible_Alarm (bell).

System Metric Monitoring.

Figure 5.5b Domain Status Section

Indicates most critical entry in each column.

Real Time Metrics.

‘top’ Metric Window

4 Future Developments.

- Distributed Network Management.

- MUTiny NEM GUI.

- Web Based Topology Display.

- An Applications Interface.