RFC Editor Reporting
August 2013
1. Monthly Summary
The following numbers represent the August 2013 statistics for documents moving
through the RFC Editor queue.
Submitted 15
Published 15
Moved to EDIT 15
Number of Documents in Queue per State at EOM (as of 1 September 2013)
EDIT 9
RFC-EDITOR 6
AUTH48 16
AUTH48-DONE 5
IANA 2
AUTH 0
IESG 1
REF 1
MISSREF 35
2. Submission and Publication Rates
While there were 23 Internet-Drafts approved for publication in October 2011,
there were an additional 25 released from MISSREF. This means that a total of
48 documents entered the EDIT queue in October. Unfortunately, due to a flaw
in the original design of our queue reporting, once the large group of MISSREF
documents were released into EDIT, the time-in-state for the EDIT state became
skewed, as the 2nd, 3rd, etc. generation MISSREFs were not originally counted as
being in MISSREF (they were mistakenly counted as being in EDIT -- this was
cleaned up a bit later).
The trend of releasing documents from MISSREF continued in the last quarter of
the year, as 10 more documents were released from MISSREF in December. So,
while the stats show that 25 documents entered the queue, 35 entered the EDIT
queue.
In May 2012, there were 29 Internet-Drafts approved for publication, but an
additional 12 were released from MISSREF and moved to EDIT. This means that 41
documents moved into EDIT during May 2012. Additionally, at least 5 documents
were released from MISSREF into EDIT during June, which means that 26 documents
were released into the EDIT queue.
Documents are continuing to be released from MISSREF, as another 11 were
released from MISSREF into the EDIT queue in July. That is, while 36 were
approved for publication, 47 documents were released into the EDIT queue. In
September 2012, no documents were released from MISSREF, but the average page
count of documents moved to EDIT increased significantly, with an average page
size of 37 pgs/doc (though average page counts throughout the year are lower –
approximately 25 pgs/doc).
While 2011 was a record high for document submissions (364), submissions in
2012 (323) slowed a bit. In both years, the RFC Editor was able to outpace the
submission rate, publishing 390 and 338 RFCs, respectively. (This is possible
because of the documents that have been released from MISSREF throughout the
year.)
In March 2013, the Independent Submissions Editor (ISE) began to use the
Datatracker to track documents submitted for consideration as an RFC. Once the
Datatracker fully met the ISE’s needs, the ISE-related documents that had not
been approved for publication were removed from the RFC Editor queue – a total
of 36 documents were removed from the queue.
1
The following table shows the average submissions and publications per month on
an annual basis:
Year Submissions Publications
2008 27 26
2009 27 24
2010 29 30
2011 30 33
2012 27 28
In the last quarter of 2010, the IAOC provided additional funding to support a
part-time editor. With this extra help, we’ve seen increased production
numbers and shorter times in RFC-Ed-timed states, and we’re seeing the RPC
keeping pace with the rate of document submissions.
There are a number of factors that affected the processing times since 2009; a
few of significant importance are noted here:
a. November 2008 – RFC 5378 was published, defining a new copyright notice
for RFCs.
• This created the “pre-5378 problem.”
• It also caused non-IETF stream documents to be put on hold, as it
did not account for the Independent Submission and IRTF streams.
b. February 2009 – An RFC 5378-fix was approved and announced, introducing
new text to serve as a work-around for people experiencing the "pre-5378
problem."
c. September 2009 – An updated TLP was announced to resolve issues
surrounding the inclusion of the BSD license in RFCs.
d. December 2009 – TLP 4.0 was announced, freeing the non-IETF stream
documents for publication.
e. December 2009 – RFC 5741 was published, defining new header and
boilerplate material for all streams.
f. November 2009-January 2010 – RFC Editor focused on transition from
USC/ISI to AMS.
g. January 2010 – RFC Editor model implemented as defined in RFC 5620.
h. March 2010 – Incumbent RSE (Bob Braden) passed baton to Transitional RSE
(Glenn Kowack).
i. March 2011 – Transitional RSE passed baton to Acting RSE (Olaf Kolkman).
j. October - December 2011 – At least 35 documents released from MISSREF
into EDIT queue.
k. December 2011 – Transitional RSE preparing to hand baton off to RSE
(Heather Flanagan).
In March 2012, the RFC Editor added time in state to the database, so that
stats would be generated directly from the database (instead of through a
multi-step, multi-file process. As such, data post March 2012 is more accurate
then previous data.
In the following pages, the graphs below show that RFC publication is typically
lower during the months of November-January, which we attribute to the holidays
and vacation season. The first set of graphs show the annual submission and
publication rates for RFCs over the past 3 years.
2
Annual Submission and Publication Rates
3
3. Queue Processing Times
The subsequent figures show the processing times of documents as they move
through the RFC Editor queue. The diagrams show document counts, page counts,
and average times in queue per state (EDIT, RFC-EDITOR, and AUTH48).
There was an increase in the size of the EDIT queue over the last quarter of
2010, as members of the RFC Production staff invested time on other
developments and because of the usual slow period at the end of the year (see
Section 2). An additional part-time editor was brought on to help with the
slowly building queue and in anticipation of the expected burst before AD
turnover.
Note that there is a ripple effect, as spikes in document and page counts may
be due to clusters of documents moving through the queue together. A cluster
does not move to the next state until the entire set is ready to be moved. You
will often see bursts in EDIT, then RFC-EDITOR, and finally PUB, as the set of
documents move through the states together to publication.
Generally speaking, the more documents there are in the queue, the longer it
takes for documents to move through the queue.
Note: the huge spike starting around week 42 in the “Time in EDIT State, Mean &
Median” figure is because of the skewed data mentioned in Section 2. The
skewed data will continue to exist until EDIT has been completed on all of
newly released MISSREF documents (i.e., they have moved to RFC-EDITOR state).
4
EDIT State 2013
5
RFC-EDITOR State 2013
6
AUTH48 State 2013
7
AUTH48-DONE State 2013
In July 2012, the RFC Editor introduced a new state with the following
definition:
AUTH48-DONE = Final approvals are complete
8
4. SLA Compliance Levels
The chart is a measure of our performance levels. The following graph shows
the total number of documents published per month and the time those documents
spent in an RFC-Editor-timed state (EDIT and RFC-EDITOR states). The goal is
to have an RFC Editor time of fewer than 6 weeks.
