Soon as I get a minute, I'll go find the code again and post it somewhere. But, I do recall that the results were pretty consistant from one run to the next, and I took the numbers from an email I sent to Si which I copied and pasted from Windows console (although for 20 instead of 2000 finds).
Again, the tests were initially just to compare SOAP and XML-RCP to get a feel for the viability of using them in production, hence the contrived nature of the "test case". Skip -----Original Message----- From: David E Jones [mailto:[EMAIL PROTECTED] Sent: Monday, September 17, 2007 2:03 AM To: [email protected] Subject: Re: Accessing database through bsh script. Are you sure about these results? The test case sounds really funny, but however it was done if the same operations are being done with both this sort of difference would REALLY surprise me. In the first 2-3 years of development on entity engine I spent around 200 hours doing profiling and optimization, and based on that the comparison to JDBC was that the entity engine generally had around a 10-20% overhead over JDBC, and generally much less (because compared to the network overhead the EE wrapper objects, especially using recycled objects, has very little overhead). It has been about 3 years since that, but I'd still be really surprised if something has changed that significantly since then. -David [EMAIL PROTECTED] wrote: > I did some simple tests to retrieve data back when I was doings comparisons > of SOAP and XML-RPC performance. This test was on the Product table with > 2000 sequential keys and then doing a binary halving to retrieve the > records, i.e. record 1000 first, followed by 500 and then 1500, etc (to > overcome database caching). > > I did this in a loop both for > delegator.findByAnd("Product",UtilMisc.toMap("partyId", client)); > > and > > r = sStatement.executeQuery("SELECT * FROM " + DATABASENAME + " where " + > FID + " = '" + client + "'"); > > These ran in separate loops, one after the other and I moved the order so > the delegrator group ran first and then second. > > The results were that it took an averate of 6 msecs to do the executeQuery > call and 23 msecs to do the delegator call. > > Skip > > -----Original Message----- > From: Jonathon -- Improov [mailto:[EMAIL PROTECTED] > Sent: Sunday, September 16, 2007 3:44 AM > To: [email protected] > Subject: Re: Accessing database through bsh script. > > > 20msec of what total? That is, what percentage is that 20msec? > > I'm assuming there is grouping and summary data in the query? > > Another important question is, how many tables, conditionals, etc? > > BUT... before you waste time answering my questions, the real question is > this. Exactly what kind > of query is Vijay trying to optimize? Always best to treat the exact problem > encountered. > > Jonathon > > [EMAIL PROTECTED] wrote: >> In my testing, there is an approximate 20msec hit using the entity engine >> over raw jdbc, depending of course on the query, is there a seca involved, >> etc. >> >> Skip >> >> -----Original Message----- >> From: vijay Si [mailto:[EMAIL PROTECTED] >> Sent: Saturday, September 15, 2007 10:28 AM >> To: [email protected] >> Subject: Re: Accessing database through bsh script. >> >> >> Jonathon >> >> It seems that other option left is to use a jdbc connection to database. > Is >> it also imperative then, that for very complex querries having >> "groupby"..."having" and other clauses it is better to use jdbc method of >> connection. >> >> How does it affect the ofbiz philosophy of architecture if i use jdbc for >> executing these querries. As right now i cannot find any ways of >> implementing those aggregate functions(sum,min,max...) through Entity > Engine >> (without creating views). Also do you have any idea of performance > behaviour >> on using jdbc connection? >> >> Thanks. >> >> >> On 9/15/07, Jonathon -- Improov <[EMAIL PROTECTED]> wrote: >>> There's no other (easy) way to query database, aside from using the > Entity >>> Engine (which was what >>> I suggested). >>> >>> If you're worried about performance hit with a huge (and unconditional) >>> cross-product, see >>> http://www.nabble.com/forum/ViewPost.jtp?post=12590081&framed=y and >>> https://issues.apache.org/jira/browse/OFBIZ-1232 >>> >>> Jonathon >>> >>> vijay Si wrote: >>>> Hi Jonathon, >>>> >>>> i cannot use the view-entity for querrying, as it would cause a >>> performance >>>> hit. Is there any other method to get the above querry. Do u know of any >>>> ways to do it through bsh script. >>>> >>>> >>>> Regards >>>> >>>> >>>> On 9/15/07, Jonathon -- Improov <[EMAIL PROTECTED]> wrote: >>>>> Use the GenericDelegator and related classes, I think. Or maybe not. I >>>>> can't recall the aggregate >>>>> functions there. >>>>> >>>>> The entity definitions do have those, though. The <view-entity>? Oh >>> yes, >>>>> the <view-entity>. >>>>> >>>>> Try using DynamicViewEntity to construct a <view-entity>. >>>>> >>>>> Jonathon >>>>> >>>>> vijay Si wrote: >>>>>> Is there any funtionality provided for using aggregate functions >>>>>> (sum,min,max..) and count() for querrying to data base through bsh. >>>>>> >>>>>> for egg: how can we write following in BSH : >>>>>> >>>>>> Select sum(order_id),sum(grand_total) from order_header where >>>>>> order_type_id="PURCHASE_ORDER"; >>>>>> >>>>>> Regards >>>>>> >>>>>> >>>>>> >>>>>> >>> ------------------------------------------------------------------------ >>>>>> No virus found in this incoming message. >>>>>> Checked by AVG Free Edition. >>>>>> Version: 7.5.487 / Virus Database: 269.13.19/1008 - Release Date: >>>>> 9/14/2007 8:59 AM >>>>> >>>>> >>>> ----------------------------------------------------------------------- - >>>> >>>> No virus found in this incoming message. >>>> Checked by AVG Free Edition. >>>> Version: 7.5.487 / Virus Database: 269.13.19/1008 - Release Date: >>> 9/14/2007 8:59 AM >>> >>> >> > >
