33
Distributed network monitoring Stefano Gridelli Co-Founder/CEO IEEE IRI 2016

NetBeez - IEEE IRI 2016

Embed Size (px)

Citation preview

Distributed network monitoring

Stefano Gridelli Co-Founder/CEO

IEEE IRI 2016

AGENDA

2

• NetBeez: Team and Company History

• Distributed Network Monitoring: Problem and Solution

• Introduction to NetBeez

• Integrations Available in NetBeez

• Future Integrations

• Closing Remarks

NETBEEZ TEAMStefano Gridelli, Co-Founder/CEO Senior Network Engineer Ms in Computer Science at the University of Milan and MBA from CMU Designed and implemented enterprise networks for mission-critical business and

healthcare operations.

Panickos Neophytou, Co-Founder/CTO PhD Computer Science at the University of Pittsburgh Researcher at the Advanced Data Management lab of the University of Pittsburgh Developed a workflow execution model for a NSF funded project for monitoring

astronomical objects and events.

Panos Vouzis, Co-Founder

PhD Computer Engineering (Lehigh University) Postdoctoral Fellow at Carnegie Mellon University Holds two patents in automation control and developed parallel processing algorithms

on FPGAs and GPUs for bioinformatics applications.

3

COMPANY HISTORYJan 2013 Team accepted into the AlphaLab startup incubator

May 2013 Received funding from Carnegie Mellon University

Jul 2013 First beta installation at eiNetwork

2014 Private beta installations

Feb 2015 Commercial launch at Networking Field Day 9

Jan 2016 NetBeez accepted into the Y Combinator accelerator

INVESTORS

4

PROBLEMNetwork Admins can’t tell if users have connectivity.

Users have to call the help-desk to report problems.

Problems happen at remote sites where there are users.

5

CONNECTIVITY IS VITALDIGITAL CONNECTEDANALOG

80’s 90’s

DATA STORAGE DATA STORAGE DATA STORAGE

ENABLER ENABLER

TOOLS TOOLS TOOLS

ENABLER

2000

6

Data Center(Monitoring server)

InternetConnection

Office

User traffic

Office

Office

Expensive Leased Line

Internet

Expensive leased lines are paired with a common Internet connection

LEGACY WAN DESIGN

7

Data Center

Internet

Office

Office

Office

CommodityInternet

Connection

Business-grade Internet lines are paired with multi-link or SD-WAN technologies

VPN tunnel

EMERGING WAN DESIGN

8

Remote Office NYC

?

?

?Remote Office

LAX

Remote Office ATL

✓ DATA CENTER

(MONITORING SERVER)

Detection and escalation of network and application issues are left to the usersLIMITED NETWORK VISIBILITY

9

NETWORK SENSOR

PLUG-AND-PLAY, SOFTWARE, AND WIRELESS

10

Data Center

BranchOffice

Synthetic tests

Internet

VPN tunnel

Continuous network connectivity and application performance testing

ACTIVE MONITORING

11

Remote Office NYC

Network Operation Center (IT headquarters)

Remote Office LAX

Remote Office ATL

NetBeez sensors proactively detect network and application issues

DISTRIBUTED MONITORING

12

OUTAGE DETECTIONNO PROBLEMS

IN PARISDEGRADED WIRELESS

IN PITTSBURGHUSERS CAN’T GET TO SALESFORCE

!

13

PERFORMANCE DATA

14

Test Data Usage

PING RTT, Packet Loss Network and service availability, uptime, and performance

DNS DNS Resolution Time DNS availability and performance

HTTP HTTP GET Time HTTP availability and performance

Traceroute Routing, RTT/Hop, Path-MTU Routing and network performance

Iperf Bandwidth, Packet Loss, Jitter Network capacity and performance, QoS validation

VoIP Mean Opinion Score VoIP call quality estimate

Speedtest Download and upload speed Internet bandwidth verification

Type Rule Usage

Up/Down IF x consecutive measurements are NULL Loss of connectivity, service down

Performance Baseline IF short_term_avg is y times long_term_avg Performance degradation issue

Performance Watermark IF short_term_avg is > z [%/ms/s] Service Level Agreement enforcement

INTEGRATIONS

15

System Usage Example Method

Network Monitoring Tools

Complementary tools based on SNMP

IBM Tivoli, Solarwinds, … SNMP Traps

Data VisualizationCorrelate alerts from

multiple systems for root cause analysis

Splunk, GrafanaSyslog, RestfulAPI,

Middleware

Outage Escalation Systems

Handle the escalation process of alerts to assure

acknowledgementPagerDuty Restful API

Collaboration ToolsAllow teams to

communicate and share documents in real-time

Slack Webhooks, Slack Bot API

Software-Defined Controllers

Enable network devices to make intelligent routing decisions

Cisco, Juniper, …Proprietary and Open API

(OpenFlow)

SNMP TRAPS

16

Integration Benefits

• Complement end-to-end and service status from NetBeez with device information from SNMP server

HTTP GET

web serverbeez

NetBeez Server

SNMP Server

ctrl_channelsnmp_get

SNMP TRAPS

17

Integration Benefits

• Complement end-to-end and service status from NetBeez with device information from SNMP polls (SNMP GET)

HTTP GET

web serverbeez

NetBeez Server

SNMP Server

ctrl_channelsnmp_get

SNMP TRAPS

18

Integration Benefits

• Complement end-to-end and service status from NetBeez with device information from SNMP server

HTTP GET

web serverbeez

NetBeez Server

SNMP Server

ctrl_channelsnmp_get

SNMP TRAPS

19

Integration Benefits

• Complement end-to-end and service status from NetBeez with device information from SNMP server

HTTP GET

web serverbeez

NetBeez Server

SNMP Server

ctrl_channelsnmp_get

SNMP TRAPS

20

Integration Benefits

• Complement end-to-end and service status from NetBeez with device information from SNMP server

HTTP GET

web serverbeez

NetBeez Server

SNMP Server

ctrl_channel

snmp_trap

snmp_get

THIRD PARTY TOOLS

21

PAGER DUTY

22

Integration Benefits

• Escalation policies based on priority levels, day of the week, time of the day, …

• Other alerting methodologies like SMS, phone calls, …

• Service grouping

PAGER DUTY

23

PAGER DUTY

24

SLACK: ALERTS FORWARDING

25

SLACK BOT

26

SPLUNK/GRAFANA

27

FUTURE INTEGRATIONS

28

https://netbeez.net/2015/08/18/closed-loop-network-monitoring/

SOFTWARE-DEFINED …

29

SDN Controller

NetBeez Server

beez A beez B

RTT(beez A, beez B, path_1) = 20 ms

SOFTWARE-DEFINED …

30

SDN Controller

NetBeez Server

beez A beez B

change_path

SOFTWARE-DEFINED …

31

SDN Controller

NetBeez Server

path_changed

SOFTWARE-DEFINED …

32

SDN Controller

NetBeez Server

RTT(beez A, beez B, path_2) = 15 ms

CLOSING REMARKS

33

Exciting time for networking

Distributed network monitoring for distributed networks

Integrations are key for distributed applications

Presentation available at: http://www.slideshare.net/netbeez