Automated WAN—Active Assurance

0:02 [Music]

0:04 active Assurance which is the monitoring

0:06 and testing of network traffic data

0:08 plane or user plane traffic often

0:11 conjures up images of expensive

0:14 appliance-based and resource intensive

0:17 equipment unscalable and

0:20 inflexible but we've broken that

0:22 Paradigm today you can visualize the

0:25 endtoend perceived performance of your

0:27 networks and services even spanning

0:30 networks that you don't directly control

0:33 on demand anywhere you like and with no

0:36 impact on the network itself let me show

0:39 you how powerful this is let's start

0:42 with testing from the dashboard we'll go

0:45 past our running monitors to our list of

0:47 tests in this example we've got tests

0:50 covering service activation Cloud

0:53 interconnect sdwan Branch connectivity

0:56 path trace and some others let's see

0:58 what it actually involves

1:01 it's actually several distinct tests

1:04 sometimes running in parallel sometimes

1:06 in sequence covering Layer Two link

1:08 paths throughput reachability and all

1:11 the way up to service response times and

1:13 error rates in layer seven applications

1:17 like the Microsoft Office 365 apps we

1:20 can see the latest test results here

1:23 showing colorcoded views of performance

1:25 against slas these tests are fully

1:28 configurable for from the dashboard you

1:31 can choose how long they run for which

1:33 subtests to include and in which order

1:36 the SLA thresholds URLs service types

1:40 and the list goes on we're going to

1:42 create a complete site activation test

1:45 in less than 10 minutes here we go we'll

1:48 test for connection integrity and

1:50 reachability connection speed Voiceover

1:53 IP and finally we'll make sure all the

1:56 Office 365 services are working properly

2:00 and we're

2:01 done that took just over six minutes now

2:05 let's save that as a template so we

2:07 never have to design the same test again

2:10 and with that done we can run the test

2:13 and see what happens let's go we open

2:17 our test template that we just designed

2:19 and now we just have to select the

2:21 client and the server now we run the

2:24 test and we can see the progress at each

2:26 step in the form of a pass or fail and

2:29 expandable progress bars which are

2:32 colorcoded to show when performance was

2:34 above or below the criteria we set here

2:37 you can see the result of our latency

2:39 test based on T wamp is a fail because

2:42 the roundtrip time was above our

2:44 threshold of 100 milliseconds from the

2:47 US headquarters to the Tokyo

2:49 headquarters the errored seconds is 60

2:53 which means it was above the threshold

2:54 for the entire one minute duration of

2:57 the test let's have a closer look look

2:59 at this rolling back up to our endtoend

3:02 latency test we can see that overall the

3:05 latency was within the threshold for all

3:08 but 1 second of the minute long test

3:11 drilling down to see this 1 second

3:13 period we can see the problem was a jump

3:15 in Jitter to 5

3:18 milliseconds so that is why the latency

3:21 test failed and we can decide whether

3:23 that's acceptable or tweak the network

3:25 to address the issue moving on we're now

3:29 testing the customer experience for

3:31 critical applications like voice which

3:34 includes load testing by sending

3:36 multiple streams of synthetic VoIP

3:38 traffic while measuring voice mean

3:41 opinion score or MOS and Office

3:44 365 by continuously sending HTTP

3:48 requests to these Services throughout

3:51 the one minute test period that was a

3:53 fairly typical site acceptance test but

3:55 you can do a lot more besides for

3:58 example you can run a Netflix speed test

4:01 run a path trace and monitor path

4:03 stability over time scan for available

4:06 Wi-Fi networks test mobile signal

4:09 strength and signal to noise ratios test

4:12 5G core Network user plane performance

4:15 through multi nodeb and uee emulation

4:18 test performance between specific qos

4:23 classes our customers use active testing

4:26 to unlock new use cases like verifying

4:29 new sites

4:30 services or changes before activating

4:33 them because they can verify reliability

4:35 and performance even if there's no

4:37 traffic in the live Network it's also

4:40 invaluable for troubleshooting because

4:42 you can quickly and easily reproduce a

4:45 scenario and then drill down into the

4:47 details to identify the root cause

4:50 instead of spending hours trolling

4:52 through thousands of performance or

4:54 fault records critical networks

4:57 supporting services like emergency

4:59 response

5:00 can be tested at regular intervals even

5:02 if not in active use giving customers

5:05 confidence that when they are needed

5:07 they'll work flawlessly now let's look

5:10 at active monitoring which like active

5:12 testing also uses synthetic traffic in

5:15 the live Network to monitor customer

5:18 experience on the user plane but this

5:20 time it's a continuous real-time process

5:23 giving you subsec granular visibility of

5:26 performance both endtoend and hot High

5:29 hop and realtime alarms that can trigger

5:32 things like closed loop remediation

5:35 through standard apis we're seeing a

5:37 summary of all monitors currently

5:39 running we're monitoring Wi-Fi OT video

5:43 voice over IP a global sdwan Cloud

5:47 connectivity and others the dashboard

5:50 automatically brings the most

5:51 problematic services to the top of the

5:53 list with a traffic light view showing

5:56 SLA compliance at a glance let's drill

6:00 down into the global sdwan and further

6:03 down into the London to Tokyo Network

6:07 everything's green so far so good now

6:10 let's look at the 24-hour view now we

6:13 see a period of around 11 minutes where

6:15 we lost connectivity going into the

6:18 graph View and zooming into those 11

6:20 minutes you can see exactly when it's

6:23 started and finished we're also

6:26 continuously monitoring latency via T

6:28 wamp and voice connectivity as well as

6:31 running ping tests 10 times per second

6:35 but crucially unlike traditional active

6:38 testing methods we're only using a few

6:41 kilobytes per second of network capacity

6:44 practically nothing this is true even if

6:47 you scale up to thousands of virtual

6:49 test agents which many of our customers

6:52 are doing today as well as endtoend we

6:55 can also visualize performance hop by

6:57 hop using path trce featur feature we

7:00 can see the roundtrip time from our test

7:02 agent to every hop along the way to the

7:05 destination address and look at every

7:07 variation of path used over time and we

7:12 can drill down to any router on the path

7:14 and see how each one is contributing to

7:17 overall reliability and performance over

7:19 time zooming right down to a few seconds

7:22 if we want it's all powered by an

7:25 elegantly simple lightweight software

7:27 solution with active insurance from

7:29 juniper Paragon you can ensure you

7:32 deploy high performance connectivity

7:34 first time every time delivering

7:37 differentiated customer experiences

7:39 throughout the service life cycle it's

7:42 easy to use flexible and scalable active

7:46 Assurance a juniper Paragon use

7:54 case