If you could start over

I am currently trying to. We are currently on 8 moving to Cloud, and we are not bring the 8 database forward.

4 Likes

I will let you know in 18 months!
The comments here have been massively helpful and we will be sure to consider them on our move to SaaS.

Currently my main query is how to clean up old data effectively… any tips would be welcomed :slight_smile:

1 Like

I think that depends on how/where you are cleaning the data. If you’re doing a re-implementation Excel will likely be your best friend.

If you’re planning on doing it in the system… DMT. Always DMT.

3 Likes

Well how about instead of cleaning, but rather deleting old data?
Is this viable / do people do this?

Just archive your old db

1 Like

Yes

I think you’d also want to get lawyers involved in this. I think there may be a mandatory time to keep data for legal reasons.

How’s that for weatherperson speak, did I leave myself enough room to wiggle out of anything?

Yes, when we went from 9.05 to 10 we had to do a re-implement because some previous person had done some SQL statements that partly corrupted the epicor database. We did get a chance to do a ton of clean up of leading zeros, Part numbering, wharehouse & bins.

Just to import that data into the “new” ERP. Keep the old DB and do SQL queries when/if ever needed.

2 Likes

Devil’s advocate - better to do bulk data changes via DMT then have them manually muck something else up during the course of maintaining their customer/supplier/part records (as that’s where it happens the most). For example, I’d rather reset buyer codes via DMT when there’s a new hire than have someone go into umpteen dozen part/part-plant records.

2 Likes

If it’s less than say, 20 items, I refuse and make them do it. I also make them fill out an Excel “Template” for the DMT they want.

2 Likes

a man in a pink shirt and tie is talking to another man in an office .

6 Likes

The idea is to type the first 5-9 characters (Type-MTL-Size1) at most and then pick. These are raw mtl so no one is typing 100x per day. But yeah we’re in-test and can renumber in minutes (plus 14 hours of DMT) :wink:

1 Like

I get to talk someone down from their random numbering idea roughly once every two years. Seems easy, except the birthday paradox exists and collision handling is way more of an issue than non-programmers tend to find intuitive. Incrementing is the robust way to go and not meaningfully different from random in practice.

No need to stop at 36! Part.PartNum is nvarchar. Unicode is supported (shhh don’t tell my users).

16 Likes

…select start

3 Likes

PartNum = Guid.NewGuid();

5 Likes

Yup.. that would be true random, and no collisions. There are THIS MANY possible part numbers using a guide (this is a 39 digit value): 340,282,366,920,938,463,463,374,607,431,770,000,000
but a guid as a base 16 value, is shorter and, looks like this: xxxxxxxx-xxxx-Mxxx-Nxxx-xxxxxxxxxxxx

I let a select few use DMT in test so they can get the template right. Then it’s an IT ticket for the Epicor Group to push it into production.

For smaller items, I have a few updatable dashboards they can paste-update to do their updates. These are more for simple updates.

2 Likes

On the other hand, unicode support might be the most effective worst way to encourage users to ditch the classic client!

12 Likes

We did not either. It runs on a VM somewhere and I hardly ever get asked to access it anymore.

2 Likes

Damn, I need to disable my BPM before I can screw with people.

4 Likes