Skip to main content

Cloud

Odd Stuff with Normalization: The + Isn’t Just a Suggestion!

I just learned something surprising today about OCS (not R2) and normalization. Eric Gleason, a friend of mine who is the OCS guru (among other responsibilities) at a customer, pointed this out to me today.

It all started when Eric and I had integrated OCS with a Dialogic and a Nortel. We originally (like 10 months ago) had OCS set up to send outside calls through the dialogic with OCS adding the 9 to get out, like this:

You dial: 555-1212

OCS sends: 9-555-1212

The gateway then routes the call to the nortel, with the 9 in front and everyone’s happy.

Yes- I know that it’s possible to configure the gateway to add the 9 for us. And usually we’ll fully normalize the numbers, including the +.

But for a reason or two, it ended up being easier to have OCS add the 9. Because the 9 was in front, I didn’t feel really good about putting the + in front of the whole thing… I felt like it would be even dumber to have OCS translate to:

+9-555-1212.

That would totally defeat the purpose of the +; it’s not an E164 number & 9 is definitely not the country code. So we didn’t have the plus there… and, like I said, things worked. This was good. No plus no foul.

All of a sudden, it stopped working. All calls to the outside were failing now. Calls to internal extensions worked fine, it was just calls to the outside. Eric and I were totally baffled – nothing had really changed on the PBX, gateway or OCS. Just all calls were failing.

We did SIP trace after SIP trace, and it showed OCS sending calls to the mediation server / gateway, but then failing.

We tried everything – re-writing normalization rules, cleaning up the routes. Nothing was working. But then one day, Eric rebuilt his laptop and forgot to update his OCS client. It was running a .64 build. And… it worked! As soon as he applied the KB rollups, it broke. He rolled back the KBs – it worked! Wha???

So Eric opened a PSS case and worked on in for a while.

The PSS guy saw the normalization rules (the ones without the + sign) and told Eric that those should have the + to conform with E164 blah blah. But I was sure that couldn’t be the problem, cause this had worked for like a year, and then some hotfix rollup came along and ruined things… so it must be a bug, right?

Nope. All the newer builds enforce the + sign now. This came as a total shocker to me. I knew that OCS liked the +, and that whenever possible, I always put the + in. But in certain cases, like with this customer, it just made more sense to not use it. Not anymore, tho. The + isn’t a polite suggestion, it’s a requirement.

Well, Eric changed all the normalization rules to have +, and it’s all working properly. I was: A) surprised that this was even the case, but B) surprised I had never seen this problem before. I know that at a couple customers , we’ve had to avoid the + for various reason on a pattern or two.

Since Eric was kind enough to pass on this info to me, I figured I’d post it as a PSA to anyone out there who may have normalization rules w/out a plus & hasn’t updated their OCS clients recently!!

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Matt McGillen

More from this Author

Follow Us