2. • What Elisa does and why should you care
• Network automation (zero touch included, no added sugar)
• Beauty benefits of graph databases
• Your guide today:
• Teemu Nykänen, Service Architect (teemu.nykanen@elisa.fi)
Introduction
3. First in Finland and in the world
1882: 1929: 1936: 1950: 1991: 1993: 2007: 2010: 2011: 2015: 2019:
Daniel Wadén
brings the
telephone to
Finland.
The telephone
network in
Helsinki is
automated.
The speaking
clock service
is introduced.
The Helsinki
Telephone
Association
builds a
telephone
network with
four
switchboards
for the
Olympic
Games.
A GSM
telephone call
is made over
the Radiolinja
network.
A GSM data
call takes
place over the
Radiolinja
network.
Thanks to the
commercial
UMTS900
network, the
3G network
can be built
quickly all over
Finland.
The 4G
network taken
into pre-
commercial
use.
The Elisa
Viihde service
combines a
modern TV
service and
the ability to
watch
recordings
using a tablet.
Elisa Internet
of Things (IoT)
service
package
launched in
Finland and
Estonia
Commercial
5G
4. Elisa key figures
2.8 million
customers, Finland and Estonia
25.4 (21.1)
customer satisfaction (NPS) 2018
Approx. 185,000
shareholders
4,800 (4,700)
Elisa employees
#1 Finland, #2 Estonia
market position
€1.83 billion (1.79)
revenue 2018
€1.95 (1.86)
comparable earnings per share (EPS) 2018
€254 million (240)
capital expenditure investments 2018
4.66 million (4.68)
mobile subscriptions, Finland and Estonia 2018
696,500 (692,300)
fixed broadband subscriptions 2018
5. Leading market position in Finland
Mobile subscriptions Fixed network subscriptions
Elisa
40%
Telia
32%
DNA
28% Elisa
35%
Telia
29%
~20 Finnet
companies
8%
Others
2%
DNA
26%
Source: FICORA
Source: Company reports Q3/2018
Post-paid market shares: Elisa 40%, Telia 34% and DNA 26%
6. Digital services for international markets
Innovative managed
services and solution
provider for large enterpise
video conferencing
Providing automation
solutions to telcos for
zero-touch network
processes
Solution for
manufacturers to enable
better decisions, reduce
downtime and improve
quality
7. The best network, the best customer experience
• We offer fast broadband connections
implemented using the best possible
technologies
• Elisa’s 4G network covers 99.8% of Finland’s
population
• Connections up to 1 Gbit/s are available to more and
more households
• We are the first in world to build a 5G network,
and soon we will offer 5G services to all Finns
• We focus on quality and on continuously
improving our customers’ user experience
• We prevent disturbances
with the help of automation
9. • SDN with Elisa twist Elisa
Software-defined Networking
• ”Facilitates network management
and enables programmatically
efficient network configuration in
order to improve network
performance and monitoring”
• Not just for fancy things
but also for devices Hybrid
networks
Codename EDN
10. • Enables automation for network. From CLI towards Intent Driven network
• Service activation/provisioning
• ”Zero touch” functionality
• Telemetry collection
• Network optimization (closed loop with telemetry data and machine learning applied )
• Multivendor (and protocol) capable. (take the power back)
• Multiple domains. IP/MPLS currently under construction
• Building things for our own needs but also planning to make parts of it commercially available
• Available from your local dealer at later in time and space
• Situational awareness required
Codename EDN
11. • Nothing to see here. Just some:
• Nodes
• Edges
• Vertices
• Connections
• Required for true automation
• Essential piece in the EDN puzzle
• Graph database fits like a glove
Network topology view (in real-time please)
12. • It’s alive! We’re in production with few services released.
• Causal cluster filled with nodes and connections
• One microservice to kill rule em’ all
• Several data streams
• Element managers
• Network discovery
• Telemetry events
• Once more, automation.
• Most of the graph is not exposed via UI
• Sharing data for analytics, anomaly detection etc.
State of a nation
14. • Around 1.3 million nodes and 1.9 million connections
• To date, we have modelled:
• Physical layer (which can also be virtual)
• Part of the logical layer (portion of the services of IP/MPLS network)
• To be continued…
So far so good
15. • Other good graph databases in the market
• We even have in-house experience of some of them
• Querying and the power of Cypher
• Not just a graph but native labelled property graph.
• Maturity
• In-house deployment
Why neo4j?
16. • Elisa is awesome
• Finns are using loads of data. Also beware of angry 5G.
• Train it with automation instead of tight leash.
• We sell things
• Blackadder episode 3 from season 2 is brilliant
• SDN is nice EDN is nicer
• Cooler than being cool is ice cold.
• Did I already mention automation?
• Know your enemy network! Store it into graph, you won’t regret it.
• Make nodes not tables
Summary
A brief introduction to the wonderful world of use case Elisa.
We shall go through what Elisa is, does and why we dream about network automation.
We’ll cover network automation and why graph database works nicely in that context.
My name is Teemu Nykänen. I’m service architect at Elisa. A software guy drifting in a sea of network automation.
Elisa’s more-than-36-year journey is characterised by making new discoveries and challenging prevailing practices.
At different stages, Elisa has been a pioneer in almost every aspect of telecommunications technology.
In this timeline you can notice at least two interesting things. First one is the automation, which has been important to us since early 19th century and another one is the ever so trendy 5G.
Some key indicators from 2017
The market leader in Finland and number 2 in Estonia.
Top in a highly competitive market.
We would like to stay there.
In addition to being “just” operator and offering wide spectrum of services to customers and corporations in Finland and Estonia. We’re also active in other areas.
We provide new digital services for international markets. For example, services related to visual communication, entertainment services and cloud-based IT services are part of our portfolio.
Our international services are strongly based on our own capabilities and our core business, as well as on carefully selected acquisitions.
The interesting part in this context is the Elisa Automate, which offers network automation services to other operators.
As a pioneer in mobile communications, we invest nearly EUR 200 million in our data networks and systems every year.
As the market leader, we are also a pioneer in new network technologies and innovations, such as 5G.
We want to be the first to provide new services, the fastest connections, and the most comprehensive coverage to our customers.
Finland Is the Mobile Data Capital of the World, which means that Finns use most mobile data per subscription.
Usage is still rapidly growing.
This requires automation.
The Black adder.
Cunning plans are always the best ones.
SDN enables network automation.
Software Defined Networks separate data plane from control plane and add application plane.
OpenSource and commercial SDN tools available
World is not black and white. Hybrid networks are here to stay.
SDN enables automation but without that automation it isn’t very usefull.
Intent driven you don’t need to know what to do, you just need to know what the end-result should be.
“Zero touch” Plug-and-play for core network.
Multivendor We will continue to work with traditional and upcoming vendors but want to keep the control in our hands. The balance needs to be right.
Great chunk of this will be commercially available. After it’s combat proven in our network.
Automation is nearly impossible, or at least very hard without situational awareness. Which means knowing what’s in the network and in which state.
Graph datamodel is a natural fit for network topology data.
We're in production. We have couple of zero-touch related services available and backlog full of new features waiting to be implemented
We have a neo4j cluster and a microservices architecture for data management.
Our target is not to expose the graph via graphical user interface the but use it for automation.
Data offers almost endless possibilities for our network management and beyond that.
The actual models, as they represent our network quite accurately, are classified
Different layers of OSI model modelled in the graph.
This allows us to create traversals for example from the port of the card to the service which might be ultimately connected to paying customer.
Around 1.3 million nodes but we’re only getting started. There’s still loads of work to do and nodes to be stored.
Graph data model is awesome and suits the needs of network topology perfectly.
But it would be useless without good tool for querying that data andCypher is just that.
It's very expressive query language and together with extension libraries it offers pretty much everything what we need in a nice package.
The data is pretty sensitive and since we an operator we have things like private data centres and private cloud.
This kind of development offers endless possibilities in many fronts, including machine learning and anomaly detection. And so far we're just scratching the surface. This is a cool domain to work with.