4 Replies Latest reply: Dec 24, 2012 2:13 PM by PradeepKPathak RSS

    Real time update in search

    PradeepKPathak
      Hi,
      Has anybody worked on real time search where user want to add new fields in UI for example: his profile new address and would like to search using same at the same time.
      The partial update also donot help to achieve real time. Having multiple query one to endeca search & another to respository will make page response slow.

      is there a better way to achieve this?

      Thanks
      Pradeep

      Edited by: Pradeep K Pathak on Dec 21, 2012 7:52 AM
        • 1. Re: Real time update in search
          Pankaj.Rajankar
          Hi Pradeep,

          Can you please elaborate more on your product stack, technology stack and the use case. Are you trying to add a new property (or dimension) to the index that was not there before OR trying to add new records to the index with properties that are searchable?

          Thanks,
          Pankaj.
          • 2. Re: Real time update in search
            PradeepKPathak
            Hi Pankaj - My product stacks are in flow like: PIM->ATG->Endeca->OMS(Sterling)
            The new attribute added in PIM flow via ATG to Endeca.
            The inventory update flow via OMS->IL(tibco)->ATG->Endeca.

            The search feature is not real time because of so many layers.
            It make sense to offer in retail domain price and inventory atleast at real time to avoid errors in customer journey as well better customer experience.

            Hence the requirement is to quicky add new attribute as well update existing one in index.
            Small size might work perfect for example desktop search.
            But I am looking big volume quick update like few hundred thousands record update every 5-10 min.

            Thanks
            Pradeep K Pathak

            Edited by: Pradeep K Pathak on Dec 22, 2012 1:33 AM
            • 3. Re: Real time update in search
              Pankaj.Rajankar
              Hello Pradeep
              It make sense to offer in retail domain price and inventory atleast at real time to avoid errors in customer journey as well better customer experience.
              This a very tall order to try to achieve. Even if you have up to the date info a scenario like this can happen: say some product has 2 counts left in stock and if 5 different users add it to their cart and keep shopping, the 3 users who check out in the last will get an error. Where as if you check stock before letting users add items to cart and hence only first 2 users are able to add to cart but then abandon the session you loose a sale.
              Hence the requirement is to quicky add new attribute as well update existing one in index.
              But I am looking big volume quick update like few hundred thousands record update every 5-10 min.
              If you are just updating some property or dimension values on a record you can think of doing the partial updates but it is not synchronous like the way you are asking for. I have heard about some WSDL based interface to update records in the latest version of Endeca but haven't explored it. Try the Endeca Information discovery forum for more on that.

              BTW, how long does the baseline take and how long do you anticipate the few thousand records to be updated in? In the end I think it all boils down to the business requirements and the constraints (ie how much time and money are you willing to spend) for your project.

              Pankaj.
              • 4. Re: Real time update in search
                PradeepKPathak
                I agree with you Pankaj that Its not achievable 100% and the kind of scenario you mentioned might be very rear.
                The goal of this solution is to avoid more checkout errors.

                Information discovery is all together another technology and cannot be leveraged to commerce.
                Partial update is taking around 1.5 Hrs to index 100K products update in my case which is major concern to get up-to-date data quickly in search page.

                Thanks
                Pradeep K Pathak