In this case query will need to traverse these 1000 rows and discard them unless you have static data and you can pre-create positions for all key lookups.
You can speed things up by using covering index – have key on (key,id) and fetch id only initially when fetch data for just those 10 rows you will display. This can be done as extra query (good if you have something like memcache) or you can have it done as a single query.
runnr
18 years ago
Hi Peter,
Although I think I understand the performance issue of using large limits, I don’t completely understand the solution you propose.. I think I have read about this somewhere.. cant really recall.. it will be great if you can shed some more light on your solution
And many thanks for doing it – learnt a lot. A video should be turning up eventually in the near future – will let you know.
The video is posted: http://video.google.com/videoplay?docid=-2461636755789321382
Hi, I’ve read the slides and have some questions about the LIMIT.
As you said in “More about LIMIT”:
Protect your application from large limits
– People may not go to page 500 but search engine bots
well may do.
If I really need to do “select a,b,c from table where key=’val’ limit 1000,10” for paging, is there any way to avoid slow query?
Pahud,
In this case query will need to traverse these 1000 rows and discard them unless you have static data and you can pre-create positions for all key lookups.
You can speed things up by using covering index – have key on (key,id) and fetch id only initially when fetch data for just those 10 rows you will display. This can be done as extra query (good if you have something like memcache) or you can have it done as a single query.
Hi Peter,
Although I think I understand the performance issue of using large limits, I don’t completely understand the solution you propose.. I think I have read about this somewhere.. cant really recall.. it will be great if you can shed some more light on your solution
Thanks!
Hello Peter,
thank you for posting the slides !
Julien