As noted here we started installing wireless gear at Naylor school. As of Friday afternoon there were just two Access Points left to go in -- we're waiting for our cabling contractor to get wires run into the main office and the cafeteria, both of which have proved to be difficult runs. There are thirty APs up and running.
Naylor is the first old school we've done, so we're trying to get a handle on how well signals propagate through the very heavy brick construction of the old part of the building. The new addition is modern steel framed construction with sheetrock walls -- the 802.11b/g and 802.11a signals just zip right through them with little power loss. We probably over-built the old building, but that's what learning exercises are all about. And, what looks like too much signal now may be just enough when the school is filled with several hundred ugly bags of mostly water (each reputed to be good for ~8dB of attenuation).
On Thursday I closeted myself at home an engaged in a great sorting-out of things. The things in question being our 2006 E-Rate funding awards, the original parts-lists for the schools where we are doing technology refreshes, and the list of know part changes that have come up between January (when we filed for funding) and now. It was something like a 3.1 pass process, and took all day.
Pass #1 was making the part substitution, which was not as simple as it might have been. For instance, Cisco has discontinued the Catalyst 3550 switch and replaced it with the Cat 3560. Swapping those is a quick ad-hoc query against the project database. But, the 3550 used traditional "big" GBICs in it's Gig ports, while the 3560 uses the newer SFP (small form-factor pluggable) GBICs. So, we have to sub them too as a secondary change. And, that drives a tertiary change in fiber-optic patch cables. GBICs have "SC" type optical connections. SFP GBICs have "LC" or "MTRJ" connectors. We buy the "LC" kind so that means ordering cables with SC connectors on one end and LC connectors on the other instead of just SC-SC cables. Figuring out those quantities would have meant, essentially, redoing the engineering work for each of the 13 sites I was looking at. Nope. I'm just going to buy a box of a 100 SC-LCs and call it a day. Beyond that, I had to swap from Nortel WiFi gear to Cisco WiFi gear, on account of Cisco having bought Nortel's supplier out from under them (best move Cisco's made in the last two years).
The resultant updated per-site information from Pass #1 went into Pass #2, where I created a new report that looks like my usual Excel-based bill of materials document (with three vendors and 13 sites, that would have been 39 BOMs, which I was not going to create by hand in Excel). It took some tinkering to get things grouping correctly in the BOMs -- the primary glitch was that I was not putting the vendor's ID into the BOM document name, so while the report lumped all the parts correctly by site, it wasn't breaking them out by vendor. It was one of those, when the light came on it was damn near blinding things. Once I stuck the vendor IDs into the BOM names, all of the parts grouped the way they should. The rest of the cosmetic formatting came together very quickly.
With a stack of freshly printed BOMs in hand, I started Pass #3. Pass #3 was where I compared the cost of the parts to the amount of money the Feds are committing per site. It's been nearly a year, and prices for things like patch cables have gone up. The Cisco WiFi gear is a bit more expensive than the Nortel gear too. I had to bring the dollars into balance by deleting things here and there (a couple rack-mount wire managers to offset the increase in cost in patch cables, for example). That got me another set of BOMs, and a couple final tinkers to get that last five bucks out of one BOM that my 'eyeball' corrections didn't fix.
Then it was forming PO requisitions, which I dumped on our admin analyst Friday morning. The reqs are in the system, and hopefully I'll have POs next week. We should start replacing equipment in February. Installing the WiFi equipment will probably wait until summer break.
I spent Friday afternoon assembling a new Squid proxy/caching server for Pathways magnet school. The machine we had placed there a couple of weeks ago continues to crash. Diags found a bad DIMM, but removing the faulty memory hasn't cured the failures. So, I'm starting from scratch with a new(er) piece of hardware (one that was in production as a Windows 2000 server until just a couple of weeks ago). The upside is that I'm building it with the latest version of Slackware Linux, which gets me support for Dell's OEM'd American Megatrends RAID card. So, the new box will have redundant disks, and much more space for caching files, than the old box. I hope to have the build done Tuesday, and the server back in the school by Wednesday.
I spoke with Asha this morning -- she's doing well. More news here.