Kevin DeGraaf’s Blog

I finally convinced the management to scrap our hideously awful brain-dead phone system and replace it with a modern Voice-over-IP system.

We purchased one Redfone foneBRIDGE2-EC device, one Rackform iServ R107 server, twenty-eight Linksys SPA942 phones, two Polycom IP4000 conference phones, five Grandstream HT286 analog adapters, and one Grandstream BT101 phone (to connect to a paging amplifier). We also consolidated our jumbled mismash of phone services (analog lines, a BRI, and a set of DIDs) into one T1 PRI connection.

Unfortunately, the management wouldn’t spring for new Ethernet wiring, so we had to piggy-back off our existing infrastructure, making it infeasible to use Power-over-Ethernet and thus requiring a fleet of local power adapters.

That aside, I have been very pleased with our new equipment. I was able to set up centralized provisioning of the Linksys and Polycom phones (using MySQL, Perl, FTP, and TFTP) without much difficulty. The Grandstream units support provisioning as well, but since they are so simple and we have so few of them, I didn’t bother.

On the software side, we’re using Asterisk (of course). That was the first decision we made, and all of the other components were evaluated based on their Asterisk interoperability. I plan to provide, in a future post, more technical details of how the system is set up. Stay tuned.

I would like to solicit feedback from my audience about two policy questions that came up during the planning and implementation of the system.

First: whether to use an automated attendant (AA) to answer the incoming calls. Under the old system, every call was supposed to be answered by a human. This policy inevitably led to dropped, missed, and/or rushed calls. It also annoyed those of us who don’t like to speak to humans unless it’s necessary; transferring calls based on a dialed extension does not and should not require using a human operator.

After strenuous internal negotiations, we convinced the management to let us set up an AA system to handle calls. Unlike the AA systems used by major corporations, ours is extremely simple: “Thank you for calling Blah. To reach the operator, dial zero. If you know your party’s extension, please dial it now. Otherwise, for blah, dial blah; for blah, dial blah, etc.”

Readers: what do you think? Should an AA be used to ensure that (1) every call is answered promptly, (2) extensions can be reached without involving a human, and (3) the human operator is far less busy and can devote more attention to each call for which the caller actually dials zero? Or, should the policy remain “every call is answered by a human”, even though that’s infeasible, inflexible, and old-fashioned?

Interestingly enough, in our discussions, the response was split neatly along gender lines. The guys voted for the AA, and the women lobbied unsuccessfully for the old system which, theoretically, involves “human contact”.

Second: what do you think of individual paging? When I started at this company, I was shocked to discover that the phone system was set up to allow any user to individually page any other user! Specifically: (1) you’d be sitting there working, (2) your phone would beep and immediately enter into an unsolicited full-duplex (two-way) speaker-phone mode, and (3) the caller would start yammering at you. I was astonished: a company actually thinks that barging in on people’s privacy like this is a good idea? What if you’re busy? What if you’re concentrating on something? What if you’re having a private/confidential conversation? In general: WTF?

Asterisk and the SPA942 phones provide for this behavior (you just set a certain SIP header prior to executing the Dial() application), but I strongly lobbied against it. Unlike with the AA debate, however, my technician and I were the only people opposed to this “feature”. Everyone else either argued in favor of it, or took a “meh, no big deal” position. One person even had the gall to suggest that our opposition stemmed from a lack of desire to assist others (like that has any bearing on ringing-vs.-barging). In general: WTF?

Enough for now…

Let’s see here… I need a new Debian server:

xen-create-image –hostname supernova –ip x.x.x.x –tar /stuff/images/tar/staging-etch-production.tar

Okay, I’m going to need, hmm (think think think) a cool terabyte of space on it:

lvcreate -n netshares -L 1T /dev/volgrp0

Total elapsed time: about two minutes (three if you count tweaking a Xen config file on the dom0 and the fstab file in the new domU).

Heck yeah, Linux!

I’ve been a computer enthusiast since 1991 and a systems administrator since 2001. In that time, I’ve had the opportunity to work/play with a wide variety of hardware and software, which has inevitably led to the formation of opinions about various vendors. There are the good (e.g. Seagate, Crucial/Micron, 3ware, Asus, Texas Instruments, Cisco), the bad (e.g. Dell, D-Link, IBM, Gateway), and the downright ugly (e.g. Microsoft, Real, SCO).

Recently, I’ve discovered a vendor of servers and server-class components that definitely belongs in the first list: SuperMicro. The name is a little dorky, but I’ve been impressed with the quality of their stuff.

At work, we have four servers using SuperMicro parts:

I personally plan on buying an SC733T chassis and a PDSLA motherboard in the near future. I’ve been meaning to build a RAID-5 media & backup server for a while now…

If you can’t or don’t want to home-build, check out Silicon Mechanics, a Seattle-based integrator of SuperMicro products. They built-and-burned the first two machines in that table. We’ve been quite pleased with them as well.

I have no stake in, nor affiliation with, either company. I’m just a satisfied customer, and since the computer industry is teeming with crappy products from crappy vendors, I wanted to take a second to recognize some exceptions.