We have a site hosted on vignette CMS 5.5 with sql server 7.0 as the backend.Currently we are using MS Search service to search the massive database.Since we are facing performance bottlenecks while searching and fetching too many records we are interested in knowing more about the thunderstone Texis search engine.
I would like to know about the following details of the search integration.
1)We have a downloaded free version of webinator.As per vignette docs we cannot use their integration application with free webinator.Is it possible to get the full texis on a trial basis so that we can test and convince our customer to use the product if the test results are satisfactory.
2)We are intersted in doing ONLY database search(which is SQL server).Has Texis performed any benchmark performance testing of Texis performance while indexing and searching an external database like sql server
3)Does the Texis engine searches its own database and the relevant data to be searched has to be replicated to Texis database from sql server periodically for this purpose?)
4)Can we perform SQL server database searches using free webinator product.If yes where can i get the relevant documantation?
1: We don't provide Texis on a trial basis, but we do have a liberal return policy.
2: Texis is its own RDBMS, records need to be imported into it before being searched. Provide the number of records, their average size, and the machine you intend to host on and we'll tell you how much load it will sustain.
> 1: We don't provide Texis on a trial basis, but we do have a liberal
return policy.
Sorry,can u elaborate or give link to what u state as a "liberal return
policy."
> 2: Texis is its own RDBMS, records need to be imported into it before
being searched. Provide the number of records, their average size, and the
machine you intend to host on and we'll tell you how much load it will
sustain.
The configurations of the hosting server is
Hardware Specification of the hosting server
Processor Type XEON 800Mhz
RAM 4GB
SQL Server version on hosting server:
Microsoft SQL Server 7.00 - 7.00.699 (Intel X86)
Standard Edition on Windows NT 4.0 (Build 1381: Service Pack 6)
We have one massive table with nearly 200,000 records.In that one column is
a text column which is fulltext enabled.The
average datalength of the Full Text enabled column is 2368.Currently for
fetching 1,50,000 records it takes nearly a minute.We wud like to have a
response time of 4-5 secs only.
> 3: Yes
Do you have details of how to index sql server database with the FREE
Webinator so that we can test it out using ASP.
If the software does not meet your needs you can return it within 30 days for a refund.
Are you really trying to return 150,000 records to the user or application? Based on your data that is 350MB of data, which will take some time to move no matter the application. Most applications only need a small subset of the records to return to the user, such as the 20 most relevant, and will be much faster You should be able to get answers in under a second. You can look at the realtime news demo on our website, which has about the same size records as you are interested in.
Free Webinator can not index a SQL server database (that was the no to number 4). The yes to 3 was that Texis searches it's own database.
Initially we need the count of records which is grouped according to the products they fall in.Since we have implemented the paging logic using a cursor in the back end we need all the records at a time to move to the appropriate record.Only 10 records are being shown in a page.Does thunderstone have a better paging method?If yes where can i find the relevant documentation?
If you search the Vortex manual for SQL you will find documentation both for the <SQL> command and <pagelinks> functions in Vortex. The SQL command takes parameters SKIP and MAX, specifically designed for paging, and the <pagelinks> function is designed to make it easier to generate the HTML for page numbering and navigation. In our experience the overwhelming majority of searches (more than 90%) are for the first page, it makes sense to only read 10 rather than 150,000 records. This is especially true if you can get the best results on the first page, for example by using Texis and LIKEP, which returns the results in relevance order.
