4. Equipment options
» Ciena 5170
› Layer 2
› RegionalAggregation or 100G NTE
– I.e. in the co-location PoPs and RNEPs
› 1 rack unit
› 4 x 100GE ‘grey’ (QSFP28) ports
– Scalable up to a 200G regional aggregation ring
› 40 x 10GE
– Access, or multiples of 10GE for lower-demand sub-rings
5. Equipment options
» Ciena 5171
› Layer 2
› RegionalAggregation
› 2 rack unit
› 2 x “Sleds”
– 2x100G coherent
– 1x 100G coherent + 1 QSFP28
– Other sleds to be launched
› 40 x 10GE
– Access, or multiples of 10GE for lower-demand sub-rings, 4 port can run at 25G
8. Equipment options: NTE
»3926m
› 6 x 1G / 10G SFP+ ports
› 2 x 100M / 1G ports
› Modular: option for embedded x86 processor
› Virtual machines for router, firewall, DPI, IDS, flow optimisation,
site-to-siteVPNs, Safe Share, performance monitoring…?
› Evaluating performance in the lab
08/11/2019 Tech 2 Tech: Future Services on Janet 8
9. Equipment options: Optical
» Ciena 6500
› To be used as mux/demux in
regions with dark fibre
› Other deployments possible for
high capacity or extended reach
– E.g. two slot module as a 100G
transponder
08/11/2019 The new Janet access infrastructure 9
10. What will this look like?
08/11/2019 The new Janet access infrastructure 10
13. Ethernet ring protection
»G.8032/Y.1344
› Less than 50ms protection switch
–(Assuming fibre ring < 1,200km!)
› Protection at layer 2, so do not need to wait for IP to
reconverge on link failure within a region
15. IP protection
»If an RNEP fails, still need to switch to the other RNEP at
layer 3
»Bidirectional Forwarding Detection (BFD)
› Rapidly detect failures that might be hidden at layer 2
–IP access may just be aVLAN on an interface that stays up
› NOC will configure on MRS routers
› Member to configure on non-MRS L3 devices
› Static routes tied to BFD, or BGP session tied to BFD.
16. Ethernet Diagnostics
»Check circuit before being brought into service
› RFC2544 /Y.1564
– Loss
– Delay
– Throughput
»Performance monitoring
› CFM /Y.1731
– Delay
– Jitter
– Loss
»Improved visibility of Netpaths through the aggregation
infrastructure
18. Scalability
»If using OSA FC, upgrades will be down to engineer time.
› New port on the 5170, re-use the NTE if it’s a 3926/3928.
› No involvement from Openreach
»Upgrades to regional aggregation rings
› Insert SFPs, connect to the OSA FC filters.
I’ll come back to this later in the presentation
High level toplogy.
Co-location in carrier neutral points of presence, mainly Openreach exchanges.
Ethernet equipment in each PoP
Layer 3 equipment only in the RNEPs.
Three example sites, one single-homed, another with resilient access, and a third connected to a sub-ring.
This is the equipment that’s going to be in each PoP. Or at least for the first few regions we’re deploying.
Performs layer 2 aggregation.
This is the equipment that’s going to be in each PoP. Or at least for the first few regions we’re deploying.
Performs layer 2 aggregation.
We will be installing an NTE at the end of every line in the new access architecture.
Visibility of the access line
Possible use for service demarcation.
In itself this is *not* intended to replace the MRS.
Layer 2 services back to the routers for IP.
Ditto for other services such as ExpressRoute.
More to come on G.8032 in a bit.
Shows how two neighbouring PoPs might be built.
4x10G backbone. Site 1 has 1G access. Site 3 has 10G access over OSA Filter Connect, so we can upgrade to 20G or more easily. Site 2 already has 20G resilient from two locations.