One of the reasons I am doing this is because I need to set the system up as a HA cluster, so not having to worry about a DB is huge plus in terms of simplicity. Note that I plan to push updates to cluster nodes from another system. The other, as I said above, is that its just one very small table. The service I am building just mirrors one table that resides on another more complex system. So the DB is there, in this case in simpledb. Butimpledb is not what you want to use to run many small queries against, if you are looking for speed, and want to avoid bankruptcy.
I have done some simple testing and it looks like it will work fine. Starting jetty from the commandline with the standard java -jar start.jar and trying to load 1MIL rows, using querySetCell does run out of heap space, but I think that is partially just a function of using CF_SQL_VARCHAR. For some reason I am getting an error with CF_SQL_CHAR. Queries on a query object with 10,000 rows is about 16ms, 100,000 about 75ms. So it looks like if I name the queries as a function of login params and use CACHEDWITHIN and limit the number of total cached queries to some reasonable number it should work fine. Mark On Jun 17, 6:12 am, Adam Haskell <[email protected]> wrote: > Honestly if you are removing and adding records just use a DB that is what > they are made to help achieve. > > That being said I've found the limits in there past to be 50k records not > based off memory footprint but more based off performance though was doing > joins. Joins are much slower/complex certainly not as simple as a single > table scan like you would be doing, so your millage will most certainly > vary. > > Adam > > On Tue, Jun 16, 2009 at 10:20 AM, mwkorver <[email protected]> wrote: > > > I have an application where I need to authenticate requests - login/ > > password pair, plus a check on remote_host. If successful, the app > > just passes back a signed string for use with Amazon S3. This is the > > only app that this server will be running. The minimal methods I need > > for this service would then be addlogin, deletelogin. > > > I anticipate something on the order of 10,000 rows on this table. > > Would it be too much to expect to be able to do this just using > > openbd's query on query capability? I was thinking just using H2, but > > realized that I really didn't need that much horsepower and I could > > keep it much simpler. > > > Anybody here have experience with how far you can go with this? what > > is a big row count for an application.query object? --~--~---------~--~----~------------~-------~--~----~ Open BlueDragon Public Mailing List http://groups.google.com/group/openbd?hl=en official site @ http://www.openbluedragon.org/ !! save a network - trim replies before posting !! -~----------~----~----~----~------~----~------~--~---
