The NAP is in the SF BAY Area. There are 26 total participants, several orders are pending.
There is now an MPLA. 8 are participating.
There is also a LA NAP. There are serveral orders pending. There is a connection to the SF NAP on the horizon. This is being done using a Complemetary Carrier Agreement.
There are now three ATM switches. There is a new one in SF proper.
Palo Alto is full, Oakland is almost full. There is a business case been presented to add more capacity via an ATM switch in San Jose. This has been approved.
The FDDI peering load is down. This is probably due to the expansion of private peering. This means that PAC*BELL is going to soon decomission the FDDI.
Transmit and Receive call aggregates from all switches at the access ports only, no trunk stats to avoid double counting.
These are sampled at 15 minute rates. We used to do 5 minute samples, but that was lost in a upgrade. This will be corrected in the next software upgrade.
http://www.pacbell.com/products/business/fastrak/networking/nap/stats for all the details.
Time Series Analysis has been added. This is a complex formula based on autoregressive-integrated moving average.
This model filters out variance due to daily traffic fluctuation and weekly fluctuations. This will allow identification of outliers (those that have more or less traffic than expected).
A plot is showed that shows that most connections are within a 4 Sigma range, but some are not. If all NAPs adopt this strategy, then there could be some correlation of events.
Using this analysis, it is possible to create an expected plot for daily usage and then compare that against the actual data determine if there is some change that is unexpected (and may be addressed as a unique event or possibly the beginning of a trend).
A new plot shoes that there was a dip during this last summer. This may be do to the large number of colleges serviced by the providers that connect to the PAC*BELL Nap.
Challenges