Loggerhead usage statistics

[Ian Clatworthy]

On 21/04/10 11:06, Ian Clatworthy wrote:
> On 21/04/10 10:04, Martin Pool wrote:
>> On 20 April 2010 17:08, Ian Clatworthy<ian.clatworthy at canonical.com>
>> wrote:
>>> With some help from spm, poolie and I have been looking at how
>>> Loggerhead is
>>> actually used "in the wild" via Launchpad. Here's the breakdown of what
>>> pages are accessed and how often:
>>>
>>> * the files and directory in the root - 29%
>>> * a particular revision - 23%
>>> * recent changes (mainline) - 17%
>>> * the files and directories in a subdirectory - 17%
>>> * annotate of a file - 12%
>>> * other changes - 1%
>>
>> I'm glad we confirmed them (and perhaps we can get further
>> confirmation) but these are not super surprising numbers to me. We
>> have a distribution across all the loggerhead functions and any of
>> them would be useful to optimize and none are really reasonable to
>> neglect.
>>
>
> Here are some updated and corrected figures:
>
> * files - 25%
> * annotate - 21%
> * revision - 15%
> * +filediff - 13%
> * download - 13%
> * no verb - 6% (maps to changes)
> * +revlog - 3%
> * changes - 3%
> * atom - 0.5%
> * diff - 0.2%

Just for the record, this data is 11k hits over a 24 hour period. It 
*excludes* 18k hits on a single URL caused by a blog post[#]. I took 
that data out because it was atypical but poolie has correctly pointed 
out that it's those sort of things we need to speed up and be capable of 
robustly handling.

Ian C.

[#] http://www.outflux.net/blog/archives/2010/02/18/data-mining-for-nx-bit/