Weired search results

  • Hi


    We have a little "search issue".


    Is there an explanation why the search term "mueller*" and "Mueller*" does produce different results?
    When I remember correctly under MySQL it did not matter!? (Now PostgreSQL) - or at least using an older Conquest/dicom.ini version it did not matter?


    While testing I am wondering why it takes conquest itsself 0 seconds for query containing 476 results and a remote dicom host like Radworks/Centricity or Clearcanvas 16s for the same query (time is taken from conquest server log)

  • Hi,


    Case sensitivity is a setting of the database engine I believe. On SQLite there is no difference.


    As for the speed, Conquest uses internally a DICOM query just like other servers do. If I query another conquest it takes 0 s as well. A Notorious issue with queries is that each returned item (record) needs to be acknowledged. So round trip delays can slow down queries very much.


    Marcel

  • Ah ok so that means when I query Conquest from another DICOM client like Radworks each querie item/result has to be acknowledged - that does explain why 400 items take 16s. (thats a crappy design!?)

  • Hi,


    That is the DICOM standard, can't change that. What network is used? Did you try to replace the used viewers with a local conquest dicom server there? It would be interesting to know if the speed is low as well.


    Marcel

  • Hi Marcel,


    Which database, Postgres oder MySQL? - I am currently investigating this and it looks like MySQL is somehow a lot faster (over 2mbit DSL VPN 780 record in 6s - postgres 28s or timeout).

  • Hi,


    I tested it with sqlite.


    The manual shows old benchmarks where mysql and postgres went compared and both were fast. It does help a lot to have the sql server local on the dicom server PC. For our highest performance ones we use indeed mysql.


    Marcel

  • Yeah its on the same machine as well.
    I cannot explain this behavior though Postgres seems to be fast when it comes to receiver and send studies, only searching seems slow!?
    I will switch back to MySQL thoufh I don't have the time to keep troubleshooting this issue and I am better in managing MySQL that postgres anyway.
    Rebuild just takes 2,5d or so, not too bad.

  • Hi,


    at home I can query 40000 records in 17s. If I try this at work it takes 175s (depending on the client), but most time is spend after the connection is closed. I guess this is symantic antivirus at work.


    Marcel

  • Thx for the hint with the 2nd network card however I have switched back to MySQL.
    Now everything is back to normal. Much higher speed than with the Postgres Setup - there must have been something wrong.
    I will stay at MySQL for now if we do not run into stability issues, if we do I have to figure out what caused the problem.

Participate now!

Don’t have an account yet? Register yourself now and be a part of our community!