Introduction to AirWave 10

AIRWAVE & GLASS
10:00 GMT | 11:00 CEST | 13:00 GST
July 31st , 2018
Albert Raja, Aruba ERT
Network Services & Switching
albert.raja@hpe.com

3
Introduction
What?
New software build that
lets us streamline code,
add performance,
clustering.
On-Premises Management
Target Audience
Who’s it For?
Large existing customers
early on.
Customers running 2 or
more AMP appliances.

4
Deployment
As of now, the plan is to deploy in two ways,
 On-Premises
 Cloud
On-Premises: It is in-house solution where either a single node or cluster
deployment could be implemented. Single node supports up to 4K devices
whereas cluster could support up to 25K devices
Cloud: The first solution is market which allows devices to be fully deployed and
managed over cloud implementation, similar as of Aruba Central

5
On-Premises Deployment Architecture
Aruba Controllers / Campus APs Aruba Switches Aruba Instant
RADIUS/TACACS Servers
Syslog Servers
SNMP Trap Receivers
Corporate Network
• Single server supports 4K devices
• Cluster supports 25K devices
Supported Infrastructure
Physical or VM AW 10 Cluster

6
Cloud Deployment Architecture
Aruba Controllers / Campus APs Aruba Switches Aruba Instant
Supported Infrastructure

7
Fundamentals
 The hardware sizing guide for on-premises deployment is remodelled for Airwave 10
 AirWave 10 will have all the features that of 8, Mobility Master (WMS), ALE and Glass (Roadmap)
 Initial release support would be for Aruba wired and wireless only
 The ISO would not be provided to Customer
 Customers with multi-vendor devices should maintain both 8 and 10 initially

8
Licensing and Migration
 AirWave 8 license would be migrated to 10
 A new support portal would be launched for this license conversion
 Migration of database AirWave 8 to 10 is reliable, waiting for more update
8 10

10
What’s New in AirWave 8?
 Device Monitoring/Management
 Code Enhancements (NGINX, HTML5, RabbitMQ)
 Database Modifications (Postgres 9.4, Redis)

11
Device Monitoring and Management
Controllers – Managing Local configuration

12
Switching – Group Template and Baseline Configuration

13

14

15
Code Enhancements
NGINX over Apache
 Light-weight resource utilization
 Load Balancing between threads allocated
 Response time during load

16
Code Enhancements
 Open-source and User-Friendly
 58% Faster than Flash in Linux
 Access of HTM5 in Mobiles/iPAD - Best Output/Content
 Stability of Flash in Linux
HTML5 over Flash

17
Code Enhancements
RabbitMQ over AirBus
 RabbitMQ is an open source message broker software
 AMQP – Advanced Message Queuing Protocol and plug in architecture
 Uses high reliable python scripting in queuing incoming and outgoing messages
 Gateways for AMQP and HTTP protocols

18
Database Modifications
Postgres
 Postgres is upgraded from 9.2 to 9.4
Redis
 Redis is an open-source in-memory remote database
 It runs in virtual memory
 Uses small structure for high durability

19
Architecture
SNMP Fetcher
PAPI Handler
Database
&
RRD
V
i
s
u
a
l
R
F
HTTPD/USER-INTERFACE
Async Logger Client
PEF Processor
Inline Stats Processor
VRF SYNC
Swarm Handler
ExecUI Query
Redis
Config Pusher/Verifier
UCC Processor
RABBIT MQ
AA- Amon Aggregator
Work
Queue
AA
AP Watcher
Service Watcher, Low Level
Service Watcher, SyncD
Controller/IAP/Switch

20
Performance & Troubleshooting
 The CLI access has been limited from 8.2
 Performance and Troubleshooting could be still be done with UI
System -> Backups
System -> Status
System -> Performance

21

22

23

24

25

26
VisualRF Statistics
https://<AMP IP>/visualrf/statistics.xml
Output:
<memory type="heap" max="3959488k" committed="3959488k" used="2347483k"/>
<service started="true" failed="false" uptime="7d:19h:27m:32s" request_time_ms="29" queue="0" queue-
completed="6740" internal="0" internal-completed="106"/>
<thread name="Message" blocked-count="1" status="TIMED_WAITING">
<service started="true" failed="false" uptime="7d:19h:27m:47s" request_time_ms="0" queue="0" queue
completed="0" internal="0" internal-completed="606944" internal-hp="0" internal-hp-
completed="922568" additional="message-bus messages count [vrf 606743 vrf_sync 922568 channel_util 0
rtls 201]"/>
<thread name="Grid Builder" blocked-count="67" status="TIMED_WAITING">

28
Why Glass?
 Glass is the fastest single point of access to monitor both wireless and wired
infrastructure
 It could be used in two ways,
Single Node
Cluster
 A single node monitors up to 30000 and 50000 devices with controller AMON and
SNMP data respectively
 Cluster monitors up to 60000 and 100000 devices with controller AMON and SNMP
data respectively

29
Why Fastest?
 Glass uses 4 major open source tools yet the best futuristic in the market,
Kubernetes
Grafana
Elastic-Search DB
Kibana
 API protocol is used in accessing the data in and out.

30
Kubernetes || Elastic-Search DB || Kibana || Grafana || API
Kubernetes: It breaks up an application into logical units called as ‘PODS’ placed in containers for easy
management
Elastic-Search DB: It is used to perform and combine many types of searches — structured, unstructured,
geo, metric — any way an application want. It is a flat DB which makes queries faster.
Kibana: It provides visualization capabilities on top of the content indexed on an Elasticsearch cluster and
logs it
Grafana: It makes it easy to customize the display properties so that the perfect Dashboard could be created
API: It is a software intermediary that allows two applications to talk to each other.
Northbound interface allows a particular component of a network to communicate with a higher-level
component.
Southbound interface allows a particular network component to communicate with a lower-level
component.

31
Architecture
Feeder and Database

32
Single Node vs Cluster
EMEA-WEBINAR @Aruba Networks a HPE Company
 A Single node should be installed in 1 TB hard drive, 16 Cores of CPU (2*8) and 96 GB of Memory
 Single Node/Cluster, the appliance should have a fully qualified domain name
 A cluster should be minimum of 3, it could not be created with 2 nodes
 Kubernetes distributes pods across 3 nodes to provide high availability in case of failure, at least 2
nodes should be up for cluster running
 Kubernetes dashboard, Grafana monitoring and Kibana logging can be accessed from Master node only

33
Install/Upgrade and Single Node/Cluster Setup
Installation:
 Hardware box comes up with Glass 1.2 installed in it and takes to setup page on power on
 In case of VM, use the OVA to deploy Core OS
 Login to the server using the default admin username and password (admin/admin)
 At the command prompt, execute the command # sudoglass-install
 After the installation completes, execute the command # glass-setup to configure network

34
Single Node and Cluster:
 The second step in # glass-setup is to set as single node or cluster
 In case of multi node cluster, virtual IP should be configured with Master and Master-HA
nodes
 Future cluster setup could be done by executing # cluster-setup command from one of the
nodes CLI. This node becomes Master Node and other 2 nodes become Master HA Nodes

35
Upgrade:
 Upgrade could be done from UI and CLI.
 UI, the option would pop-up for upgrade

36
Upgrade:
Login to the single or master node with the admin username and password
 At the command prompt, execute the command # glass-upgrade
 Run the CoreOS and then Kubernetes upgrade followed by it
 Running either upgrade on cluster Master node will push to all Glass in cluster
 A reboot is necessary after both the upgrades respectively

QUESTIONS?
Email: albert.raja@hpe.com
Phone: +1 281 257 7820 Extn.: 6342447
Time: US – Covering EST and PST

Introduction to AirWave 10

More Related Content

What's hot

Similar to Introduction to AirWave 10

More from Aruba, a Hewlett Packard Enterprise company

Recently uploaded

In this document

Introduction to AirWave 10