operation of vdle concept revised c&w proposal. background c&w continue to believe that in...
TRANSCRIPT
Operation of vDLE concept
Revised C&W proposal
Background
• C&W continue to believe that in the “virtualised” situation CPs should pay connection & rental charges as if they would have in the “actual” situation.
• Our proposal has been amended to reflect industry feedback– Concerns around complexity– Concerns around wishing to allow for overflow
Proposal
• In the situation where a DLE is closed, all payments will be based upon virtualised concept– No port rentals on actual ports will be paid, and with
one exception this will also include no port rentals on actual NGS capacity
• Number of virtual ports to be paid for at vDLEs and the NGS to be calculated according to measured monthly traffic– By reference to Table – see next slide
Port Requirement TableMonthly minutes up to (Equivalent Erlangs) Port Requirements
300,000 19.8 1683,333 45.1 21,072,727 70.8 31,460,606 96.4 41,851,515 122.2 52,243,939 148.1 62,636,363 174 73,030,303 200 83,425,757 226.1 93,819,696 252.1 104,215,151 278.2 114,610,606 304.3 125,007,575 330.5 135,403,030 356.6 145,800,000 382.8 156,196,969 409 166,592,424 435.1 176,989,393 461.3 187,386,363 487.5 197,783,333 513.7 208,181,818 540 218,578,787 566.2 228,975,757 592.4 239,372,727 618.6 249,771,212 644.9 2510,168,181 671.1 26
Assumptions in Table Derivation
• Minutes Erlangs– 9 network busy hours per day– 5 days a week– Total weekend traffic = approx 20% total weekday
traffic– 4.33 weeks/month– 5% overhead allowed for discrepancies to the above
assumptions, differing month-lengths etc
• Erlang Port Requirements– Standard TC4 dimensioning tables
Worked Example
A250k mins
B450k mins
C650k mins
D750k mins
NGSOLO transit300k mins
Total actual traffic = 2.4M minsi.e. approx 160eActual ports = 7xE1
CP
C&D are connected by vDLE, A&B are not
vPort/VIEC requirements:A – not applicableB – not applicableC – 650k = 2 x E1D – 750k = 3 x E1NGS – (250k+450k+300k) = 1M = 3 x E1
CP does not pay for the 7xE1 actually used on NGS route as they are paying for all virtual capacity (including NGS)
Exception : if CP consistently over-dimensions NGS route (e.g. has 9xE1) they would pay the additional actual port rentals
What if traffic exceeds capacity?• Once per ACO period, measure traffic actually carried versus the vDLE
capacity agreed• If capacity is insufficient, CP to have two options;
– Option 1 : right-size route• CP pays for ports required retrospectively• To discourage routine under-dimensioning, 50% penalty to be paid on these ports
– E.g. if 2 additional virtual ports were required, for retrospection period 3 ports will be charged for, but thereafter the penalty would not apply
– Option 2 : pay for overflow• CP pays for any additional minutes over those allowed in Table at ST daytime rate• Overflow assumed to be same in all months
– ie measurement would only take place every 4 months, so all 4 months would have same surcharge applied
• Accounting on bulk basis (e.g. “50k mins overflow = £52 surcharge payable”) rather than on call by call
– No impact into core EBC charging
• For avoidance of doubt both options would be available to CPs
What about during the transition period?
• Given BT may adopt a pepper-pot approach to migration, what about locations where there is a mixture of migrated/non-migrated nodes?– Eg at a given time during migration in example A and C could be
migrated but not B and D
• C&W propose to treat the location as migrated and pay as per virtualised rules once the first CP-connected DLE has been migrated in an area– This is necessary because it is complex to mix virtual and actual
capacity charging– From this point, actual capacity charges to outstanding DLEs would not
be levied• Eg D would be charged on a virtual basis, regardless of physical capacity
still being in place
Why aren’t other options acceptable to C&W?
• Partitioned routes– Extremely inefficient for C&W, particularly as multiple C&W
nodes are interconnected to each BT node
• Simple “divide minutes by xxx to get number of E1s”– This does not take account of dimensioning efficiency
• Eg a 5xE1 route will carry 6.2 times as much traffic as a 1xE1 route – simplistic formula would assume 5 times!
• As C&W’s routes are large, this approach would require approx 30% more virtual ports than actual ports used today
• Call by call accounting in EBC– Would require changes to EBC itself rather than (reasonably)
simple periodic reconciliation