as I understand it, each of the services (classification, pre-processing of the image, OCR, Field data extraction and export are all essentially separate services/queues within the Epicor IDC Server service and each is managed independently by the server service and a queuing app it installs called RabbitMQ. (check your program files directory).
Knowing that, we got into a conversation with Ancora guys and learned that the pre-processing of the image is what will kill a server - depending on what you are doing. For example, if you check all the boxes for de-speckle, orientation, rotation, blank removal, vector extraction, and graphics cleanup - you are seriously adding time to the loop for each and every document. And then there is document separation… and did you see this little suggestion?
On the ECM side -
Are you doing any OCR pre/post work on the ECM side also? I’m told the ECM OCR stuff is even slower because it has to farm out all the work to the active/registered ECM clients installed on the network. Yes, even though the import is processed at the server (folder watching) the processing is actually done in a distributed fashion by registered ECM Clients. We were told that if it got bad, we could install just that client piece on a few of our dat center servers and get better performance.
I unchecked the 3 checkboxes, just for PO’s so as not to interrupt our Payables team.
What happens is a single PO document (or other batch type) gets behind a half dozen or more Invoice batches all with “OCR performed - ready for Capture” status. Two batches will be “In Data Capture” at any given time.
Each Invoice Batch has a max of 5 documents (some have multiple pages but never hundreds) and each batch takes about 5 minutes in the “In Data Capture” step.
This means a single document waiting behind 6 batches is still waiting 30 minutes to get into Data Capture step and this is not acceptable.
Yeah, I see what you are saying and I think you are experiencing what I was talking about, albeit larger scale. I don’t have stats for multiple document types X number of vendors X number of documents X (anything else particular about your setup like calculable fields in the DFD, validation formulas, etc. that might all be part of the problem)
Do you have any of those in the DFD that are doing look ups back to Epicor? You may want to consider doing some work in SQL to make those look ups local to IDC (through some SQL replication/synching) or something.
Keep after them and please let us know what comes of this. I am very curious as I’m about to roll out to 3 more companies by year end.
We heard back from Epicor and Ancora today. We need a few hardware changes and a version upgrade (we were implemented on 7.2 earlier this year and were not told of a newer version). This will accommodate our approx 20,000 docs per month. This will be done tomorrow.
We don’t use Epicor. We don’t have IDC connected to our ERP system.
Sorry - the posting here usually indicates ERP as well - but you are welcome none the less!!
Look forward to your results improving! I think a number of us would be interested in the particular specifications of the changes you are making once you’ve tried them out. This group is pretty technical and we enjoy the gory details
@Barb, like you, I have documents flowing from DocStar (Epicor IDC => Epicor ECM =>) into non-Epicor ERP system. One process has a limited # of AP invoices going into Dynamics GP. Another process has customer POs (for pickup services) going into a 3rd party transportation system as our Sales Orders.
@Barb, you might not want to share which ERP system you’re integrating with DocStar. But if it’s Dynamics GP and the 20,000 docs per month are AP invoices, I would love to ask you a few questions privately. I’m on the verge of pursing AP Automation for almost all of our AP invoices. So my questions would involve how you’re addressing some things with the Coding (GL distributions) and AP Batching.
Yes, we are using Dynamics GP and most of our docs are vendor invoices. Thanks for the upgrade notes. We will be working directly with Ancora to ugprade.
I believe the default had those unchecked and I just left it that way. Our first consultant set the batch size to one file and I have done our other imports the same way.
I have separate IDC and ECM servers based on a DocStar recommendation and I am processing no where near 20,000 documents a month. My IDC VM is 4 cpu and 16 GB, ECM is 4 cpu and 24GB.
I was curious about my timing so I did a batch of 38 POs and they finished in 12 minutes in IDC. The first one was ready to verify in 2 minutes, so I could have started verifying. We also went live on 7.x and I upgraded to 9.11 in March and am planning on 9.22 as soon as I can get time.
there is a nice ‘what’s new’ doc in the EpicWeb download section. For IDC 9.22 there are some cool new things - download images from a batch, new formula stuff, misc. performance changes.
“Better, Faster, Stronger”. I hope the person in the booth who provided you with that insightful information at least gave you some swag, perhaps an ancora t-shirt? Because clearly that person had no useful insight at all.
@gpayne, thank you for the direct link to the IDC 9-2 release notes! My professional services partner only shares such release notes when he asks to schedule the upgrade. I prefer to review the release notes well in advance.
Auto-Save is an interesting concept. Glad it’s an optional feature. Shouldn’t ancora first address the lack of a “You have unsaved changes – do you really want to close without saving?” notification when an IDC user in Data Verification attempts to return a batch with changes? Or is that too old-school?
