I microbenchmarked the storage queue standalone. i.e. the set and clearrange mutations
were performed solely at the in-memory storage queue. No other FDB components were involved
in this test. And hence the numbers presented here the best case numbers.
Test setup:
- 100M mutations: about 5% clearRange and 95% set mutations
- 100M rangeReads
- Keys/Values generated using deterministicRandom()
- A new version generated for each mutation (i.e. it's an extreme version test)
Performance comparison between std::map and std::deque for VersionedMap"
std::map std::deque
Time to perform the mutations 220.066 218.784
Time to perform buffered readRange 184.423 171.578
Since storage queue nodes account for a large portion of memory usage,
we can save space by only allocating 96 bytes instead of 128 bytes for
each node.
Remove the use of relative paths. A header at foo/bar.h could be included by
files under foo/ with "bar.h", but would be included everywhere else as
"foo/bar.h". Adjust so that every include references such a header with the
latter form.
Signed-off-by: Robert Escriva <rescriva@dropbox.com>
* Detail names now all start with an uppercase character and contain no underscores. Ideally these should be head-first camel case, though that was harder to check.
* Type names have the same rules, except they allow one underscore (to support a usage pattern Context_Type). The first character after the underscore is also uppercase.
* Use seconds instead of milliseconds in details.
Added a check when events are logged in simulation that logs a message to stderr if the first two rules above aren't followed.
This probably doesn't address every instance of the above problems, but all of the events I was able to hit in simulation pass the check.
sfc-gh-tclinkenbeard