BIND 10 #3346: Revisit logging design

[BIND 10 Development]

#3346: Revisit logging design
---------------------------------------+----------------------------
                   Reporter:  shane    |                 Owner:
                       Type:  task     |                Status:  new
                   Priority:  medium   |             Milestone:
                  Component:  logging  |              Keywords:
               CVSS Scoring:           |             Sensitive:  0
            Defect Severity:  N/A      |           Sub-Project:  DNS
Feature Depending on Ticket:           |  Estimated Difficulty:  0
        Add Hours to Ticket:  0        |           Total Hours:  0
                  Internal?:  0        |
---------------------------------------+----------------------------
 This ticket documents a proposed synthetic benchmarking for research
 purposes.


 We use log4cplus where each process logs directly. This was chosen for
 performance reasons, but we have had a lot of problems getting logging to
 be free of concurrency problems, in spite of our extra-careful locking.

 We should benchmark an approach that sends data to a centralized logging
 task to see what the performance implications are.

 Ideally we should check a few different secondary storage approaches:

 1. Normal spinning disks
 2. Solid-state disks (SSD)
 3. RAM disk (for best-case/worst-case measurement)

 We should also check with no concurrent logging, mild concurrent logging,
 and massively concurrent logging.

-- 
Ticket URL: <http://bind10.isc.org/ticket/3346>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development