Improvements in Customer Empowered Fibre Networks Stanislav Šíma CESNET 4th CEF Networks workshop Prague 2007.

Improvements in Customer Empowered Fibre NetworksStanislav Šíma

CESNET

www.ces.net

4th CEF Networks workshop Prague 2007

4th CEF Networks workshop

2Improvements in Customer Empowered Fibre Networks

Author participates on Optical networks activity of CESNET research program,

GN2 project and Phosphorus project

Presented ideas do not necessarily reflect an official opinion of CESNET or any other institution or project.



Why funding of RENs Reading of report “Digital Prosperity” is recommended

(to better understand the nature of the new economy, innovation, productivity and broad-based prosperity of country) http://www.itif.org/files/digital_prosperity.pdf

Research networks are an important source of Internet innovation and contribute to information technology innovation and deployment

Support of research networks development should be understood as priority in each country

http://www.itif.org/files/digital_prosperity.pdf

http://www.itif.org/files/digital_prosperity.pdf



How to improve networks Fibres (including short cross border fibres and first mile)

dedicated to research network are crucial Single fibre lines should be used, if improve cost effectiveness Pure optical networking improves functionality and economy

(photonic switches, multicast, waves transiting nodes, alien waves, resource dedication, etc.)

Use full power of photonic devices and programmability - big improvement of functionality, speed, cost and/or other parameters

Experimental Facilities (EF) including long distance dark fibres are inevitable for research of WAN improvements

Lit fibre service simplifies empowering of fibre networks Fibre network cost model is needed



Avoid slow innovation and high costs Slow innovation and high costs are in contradiction with research

networking mission Requirements (concerning research networks design, deployment,

operation and innovation during network life cycle) are substantially improved:

reduce dependency on – data transmission services: use also dark fibres– traditional network equipment vendors: use also open photonic devices for

transmission and switching Reducing dependency brings advantages in procurement and

contracting: competitive offers are accepted only Pre-procurement with electronic and photonic industry (including

collaboration on proposals of standards) will help



Network Layers in research Knowledge of research networks staff:

– Many experts: IP service, routing and switching using equipment of traditional vendors, protocols, SONET/SDH, Ethernet, telco services

– Few experts: Fibre acquirement, Fibre lighting, HW design, FPGA programming, transmission technology

Result: research programs are suggested mostly in the first field Innovation point of view is the opposite: the most important

improvements is coming from photonics technology now Economy point of view is also the opposite: about 80% of costs of

operation are transmission costs as sum of fibre usage cost (about 55%) and fibre lighting costs (about 25% ) – see next slides



Improvements of research possibilities(an example based on Cesnet experience)

Steps to reduce dependency on telco operator services since 1999 Reasons

– Strategic and technical limitations resulting from telco policy– Economics (costs exhausted too big part of funds allocated for research)

Steps to reduce dependency on traditional networking equipment vendors since 2002

Reasons– Innovation policy of vendors sometimes limits research networks development – Lack of cost-effective up-to-date photonic equipment suitable for NRENs

Diversification of funding – participation on internationally funded projects (since 2000)

Technology transfer from research to equipment production and to network services (for example by licensing) since 2006– CLA and COMBO production– Lit fibre service



Fibre network cost model General cost model was elaborated in Porta Optica Study FP6

SSA http://www.porta-optica.org/ Model is based on annualized network costs The most important task of REN from economical point of view

is usually to overcome distances Corresponding cost category is transmission cost as sum of

– fibre cost and – fibre lighting costs.

Model allows comparison of different solutions See example in attachment and Deliverable D3.2: „Economical

analysis, dark fibre usage cost model and model of operations“ for details (draft is available now)

http://www.porta-optica.org/




Advances of photonics in fibre pair lighting (kE)

Fibre pair Traditional vendors 100 km

Photonic vendors120 km

Traditional vendors 200 km

Photonic vendors200 km

Photonic vendors4x100 km

1 x GE 24.1 1.1 39.5 13.2 26.9

4 x GE 33.5 5.7 48.9 28.3 39.2

8 x GE 42.5 13.2 64.1 42.1 49.1

1 x 10GE 58.4 26.1 80.8 35.8 52.3

4 x 10GE 120.2 55.7 142.6 73.2 106.7

8 x 10GE 202.6 104.3 225.0 111.8 135.7

16 x 10GE 379.9 183.3 414.8 190.8 214.7

Based on offers received in procurements in last months in POS





Devices for open lighting Photonics industry products readily available in the market and

based on standards allow building of fully programmable optical network platforms.

This extends well-known open approach from software development to hardware and optical devices development and management.

CzechLight (CL) family is affordable set of photonic devices enabling to use full advantage of up-to-date products of photonic industry by programmable devices

Very cost effective, comparing to traditional approach Availability of transmission parameters to monitoring and

management (impairment monitoring and lightpath switching and restoration, testing of reconfigurable optical transport systems, etc.)

Fully open to improvements during network life cycle



CzechLight family CzechLight (CL) family:

– CLA: family of EDFA amplifiers– CLR: Raman amplifier, – CLS: Optical switch, – CLC: tuneable CD compensator, – CLM: multicast switch– next CL devices are prepared

Concept was proved by deployment in fibre reels, CESNET Experimental Facility (field fibres) and CESNET2 network (in-service), devices are very reliable

Development of UNIX-based management of photonic components has good results (using experience with open software development in universities and RENs)

Possibilities of shaping the future (and further development of open lighting concept)



Main types and indicative prices of CLA Main types without gain flattening for up to 8 channels:

– CLA PB01 (8900 EUR) Low noise preamplifier and booster amplifier with output power up to 20dBm and optional ALS (Automatic Laser Shutdown)

– CLA PB02 (12950 EUR) Low noise preamplifier and high-power booster amplifier with output power up to 27dBm and built-in ALS

– CLA DI01 (8300 EUR) Dual in-line amplifier with output powers up to 15dBm

Main types with gain flattening for up to 32 channels:– CLA PB01F (10870 EUR) Gain flattened low noise preamplifier and

booster amplifier with output power up to 20dBm and optional ALS– CLA PB02F (14980 EUR) Gain flattened low noise preamplifier and high-

power booster amplifier with output power up to 27dBm and built-in ALS– CLA DI01F (9350 EUR) Gain flattened dual in-line amplifier with output

powers up to 15 dBmPrices for 1-2 pc in September 2007



Dark Fibre and photonics Experimental Facilities

Open dark fibre and photonic Experimental Facility (EF) should be dedicated for support of designing the Internet of the future at all network levels (from dark fibre lighting to network applications in different research areas). – Facilities or testbeds based on lambda connections without full access to dark fibres

and programming of lighting devices are considered not sufficient for the future. The EF enables fundamental shift from current networking technologies that

do not offer programmability and flexibility required by the research community.

We distinguish between EF and testbeds (EF is multi-purpose, long term and less technology dependent)

EF enables advances in network development by supporting experiments and applications

Dilemma for EF and testbeds– real users but little innovation– networking innovation but no real users



CESNET EF CESNET Experimental Facility (EF) consists of dark fibres,

CLAs and switches EF is used for

– testing of new CL devices and new photonic products, – for bulding and operation of testbeds (now Phosphorus testbed), – disruptive experiments with new services and products etc. before

deployment in CESNET2– support of experiments with new applications and research collaboration– first mile solutions testing

EF users: we collaborate with GLIF users in the Czech Republic – collaboration of facilities users looks very reasonable

We are searching for international partners and projects with similar approach



We search long term collaborationCharles bridge built in 1357



Acknowledgement

All partners from CEF Networks and GLIF community, especially Jan Gruntorád and colleagues Lada Altmannová, Miroslav Karásek, Martin Míchal, Václav Novák, Jan Radil and Josef Vojtěch from CESNET

Above colleagues are not responsible for any my mistake .

Slides for off-line reading follow ...........



Fibre network cost modelGeneral model elaborated in Porta Optica Study http://www.porta-optica.org/

Item Cost Category Annualized costs (kE/y) Percent

1 Transmission 1676,38 69,22

2 PoP Equipment 110,40 4,56

3 Other operational 423,39 17,48

4 Advanced research 211,70 8,74

TOTAL 2421,87 100,00%

First three items are costs of operation. See next pages for examples of PoP equipment cost table and transmission cost tables.

Model is based on annualized network costs. The most important task of REN from economical point of view is to overcome distances. Corresponding cost category is transmission cost as summary of fibre cost and fibre lighting costs. Following is „Overall NREN budget table“ for Latvia as one case study for example.





Fibre cost per meter and yearFibre budget table - example

Fibre budget table

PoP A PoP B Fibre length (km) Use time (year) Fibre acquiring

cost (kE)Fibre cost per

year (kE/y)

Fibre cost per meter and year (E/m/y)

Contract type

Riga Daugavpils 230 3 910,8 304 1,32 lease

Daugavpils Rezekne 130 3 514,8 172 1,32 lease

Riga Jelgava 50 3 198 66 1,32 lease

Jelgava Liepaja 180 3 712,8 238 1,32 lease

Riga Ventspils 200 3 792 264 1,32 lease

Riga Valka 205 3 811,8 271 1,32 lease

Jelgava Joniškis 40 15 3600 240 6,00 purchase

T o t a l 1035 7 540 1 553 1,50

Annotations:kE/y – kiloEUR/year, E/m/y – EUR/meter/yearUse time is expected duration of usageContract type – IRU, lease, single, pair, other Fibre maintenance is included in fibre cost



PoP equipment cost per year budget table example

PoP equipment budget table

Type of PoP Number of PoPs

R&SE cost per PoP

(kE)

R&SE cost (kE)

Use time(year)

R&SE cost per year

(kE/y)

R&SE&M cost (kE/y) COLO cost (kE/y)

PoP cost per year (kE/y)

Border 2 130 260 5 52,00 57,20 2,60 59,80

Backbone 2 42 84 5 16,80 18,48 0,84 19,32

10G capable 2 23 46 5 9,20 10,12 0,46 10,58

1G capable 15 6 90 5 18,00 19,80 0,90 20,70

T o t a l 21 201 480 96,00 105,60 4,80 110,40

Annotations:Type of PoP: common name for PoPs using the same equipment and having the same collocation conditionsR&SE - Routing and Switching Equipment including racks, energy backups, servers, workstations, PCs, testers, etc. for monitoring and maintenanceCOLO cost: Total cost of collocation of PoPs of given type. R&SE&M cost: R&SE annual costs plus maintenance cost



Transmission costs per meter per yearTraditional lighting example

Lighting budget table

PoP A PoP BNumber

Of lambdas

Number of spans

TE costs (kE)

Use time (year)

TE costs per year (kE/y)

TE&M costs

per year (kE/y)

IL COLO cost (kE/y)

Lighting per year

(kE/y)

Lighting per m (E/m/y)

Tr. cost per m per y

(E/m/y)

Riga Daugavpils 2 7 612 5 122 135 0 135 0,59 1,91

Daugavp. Rezekne 1 1 39 5 8 9 0 9 0,07 1,39

Riga Jelgava -Joniskis 3 2 265 5 53 58 0 58 0,65 1,97

Jelgava Liepaja 1 3 114 5 23 25 0 25 0,14 1,46

Riga Ventspils 2 4 610 5 122 134 0 134 0,67 1,99

Riga Valka 4 4 421 5 84 93 0 93 0,45 1,77

T o t a l 11 21 2 060 411,96 453,16 0,00 453,16 - -

Annotations:TE - Transmission Equipment, TE&M - TE and Maintenance IL COLO costs – Cost for collocation of In-Line transmission equipment, if not included in fibre acquiring costs, Lighting cost = TE&M cost + IL COLO costTr. cost (transmission cost) per year and meter = Lighting cost per year and meter + fibre cost per year and meter



Transmission costs per meter per yearPhotonic lighting example

Lighting budget table

PoP A PoP B Number oflambdas

Number of spans TE costs (kE) Use time

(year)TE costs

per year (kE/y)TE&M costs

per year (kE/y)IL COLO

cost (kE/y)Lighting

per year (kE/y)

Lightingper m per y

(E/m/y)

Tr. Cost per m per y

(E/m/y)

Riga Daugavpils 2 7 119 3 40 44 0,00 44 0,19 1,51

Daugavp. Rezekne 1 1 9 3 3 3 0,00 3 0,03 1,35

Riga Jelgava 2 1 17 3 6 6 0,00 6 0,12 1,44

Jelgava Liepaja 1 3 51 3 17 19 0,00 19 0,10 1,42

Riga Ventspils 2 4 68 3 23 25 0,00 25 0,12 1,44

Riga Valka 2 4 68 3 23 25 0,00 25 0,12 1,44

Jelgava Joniškis 1 1 17 15 1 1 0,00 1 0,03 6,03

T o t a l 11 21 349 112 123 0 123 - -

Improvements in Customer Empowered Fibre Networks Stanislav Šíma CESNET 4th CEF Networks workshop Prague 2007.

Documents