my so-called blog (redux)

so, the new thing that’s been making the rounds lately is the idea of a blade-based PC enviroment. basically, what you have is a blade server sitting in your computer room, and these little terminal-type boxes that sit on the client’s desks and contain the hookups for the monitor/etc. it’s a new spin on an old idea: most recently, using something like a SunRay device or a Windows-based terminal, which themselves aren’t too much different from just deploying old-style X terminals or green-screen dumb terminals (to grossly oversimplify the situation) in your organization. the spin, however, is that each one of your terminals has its own PC behind it. (or, at least that’s how it seems - ClearCube’s website doesn’t seem to state whether or not their solution allows you to aggregate your blades together and carve out slices from the whole for your users. that’d be a big thing, IMHO - given the option of building a new server room to handle several racks worth of blades to service my, for example, 2,000 user organization, I think I’d opt for client desktops.)

I’m personally not seeing the point. I do agree that you’ll reduce certain kinds of support costs if you move to this setup - you won’t have to visit the client every time a hard drive dies or they need more RAM or whathaveyou, since you can just pull the blade out of the rack to do that. however, you’ll have to hit the client to see what the terminal is doing when the management software can’t contact it. (my bet on that is it’d be almost a wash, with the terminals coming out a bit ahead.) I don’t quite see how this makes your management much easier. with technologies like Microsoft’s Active Directory, you can apply policy to the machines that are fairly strict anyway. (it’s even fairly simple, albeit a bit tedious, under Server 2003.) furthermore, with stuff like Dell’s OpenManage, HP OpenView, and the various other products that are available to do this kind of thing, you can query status on the physical hardware and modify PC-level stuff remotely as well. (Dell OpenManage can let you configure the BIOS remotely, etc. etc. etc. I’m not too sure about what OpenView can do, though, since it’s a huge product and I haven’t worked with it at all.) still, unless the computer is completely and totally locked up, you can do most of this stuff remotely or with just a call to the client to say “hey, hold down that power button for about 10 seconds”. I’m furthermore interested in how you get increased uptime with a blade-based system. this is purely anecdotal, but I can’t remember the last time I had to RMA a part on my business desktop - or any of the ones in my immediate area, and I can only remember once or twice having to RMA a system-critical part on a desktop. what I do see fairly often, though, is people loading their machines down with enough adware and spyware that their machine essentially “breaks”, and I don’t see how a blade enviroment will fix this.

so, basically, what I’m getting at is that I don’t quite see the advantages of running a blade PC enviroment. feedback appreciated.

in other news, I recently attended a pretty high-level Windows Server 2003 training course and left it actually being impressed with a Microsoft product (and, believe me, that doesn’t happen very often). it also makes the Apple fanboi in me sad, since they don’t have anything near what MS has in Active Directory. but I do love how I sat down the other day and spend five minutes installing RIS on a really crappy server and then having it ready to image machines with Windows XP SP2. that was cool.

for what it’s worth, pocketmac seems to be a ton faster on my Mac mini than on my powerbook. don’t understand why but there it is.

so I’m giving the whole “information appliance convergence” thing and have thusly bought one of these (the h6315 to be exact). after playing with it for a few days I’m left with an overwhelming uncaring feeling about the whole idea - it’s great that I can sync it with my computer and have real data on the screen that’s easier to get to than it is on a standard cell phone, but on the other hand I keep hitting the start menu and firing up things if I talk on it without the headset. the iPaq has a great hand presence (better than my Axim x50, even) but it’s just too goddamned big for a cellphone. I’ll do a better writeup on it once I get a little more time with it. I have two weeks to screw with it.

basically, in the day I’ve had it, I do really like the always-on internet connectivity. when wifi is not present it’ll fall back to GPRS. and, as far as it goes, it’s pretty fast - the processor on it is a TI OMAP 168MHz model, which pales in comparison to the 520MHz XScale in my Axim, but I really can’t tell the difference. IE is slower. sync (through PocketMac, not Active Sync - don’t have a PC right now to try it with) is ungodly slow but that may just be pocketmac. (for some reason it’s way faster over BT.) connectivity rocks, as it has GSM/GPRS, 802.11b and bluetooth 1 built in. the thumboard is nice but the keys are chicklet-style and hard to type with as there are very noticeable gaps in between. it’s hard to dial using the touchscreen. I love being able to hook it into the tape adapter in my car (which I usually use for my iPod) and being able to field calls through my car stereo. sound quality is awesome - it’s the first time I’ve had a headset that I could understand the other person on fully, and I finally have a GSM phone that beats the CDMA Cellular South free phones I’ve had. that said, I bought it yesterday, and I’m resetting it and exchanging it today since the wifi doesn’t work at home (which uses 40-bit WEP - mainly to keep casual people out, natch, and not for security). I’m, however, still up in the air about full-sized PDA phones, and am not convinced yet that they’re a good idea. I need to check the return policy at expansys and see if I can try out some i-mate models (the i-mate Jam PDA phone and i-mate SP3 smartphone) and maybe the Symbian-based Sendo X. the Jam lacks 802.11 but it does have an SDIO slot, and it’s evidently quite a bit smaller than the iPaq (around the size of the Motorola MPx220, which is somewhat bigger but not amazingly so than my V180 that I hate with a passion). (FWIW, built-in wifi is a pain in the ass when it’s on, you’re driving, and you’re trying to make a call - every few seconds it wants to know which of the myriad wifi networks that are around you at any given point in time connect to the internet. and no, the h6315 doesn’t have that cool dial-by-voice-command feature the i-mate PDA2K/audiovox somethingorother has.)

as an aside, a few cool things about OS X: Cmd+Ctrl+D while you hover over a word brings up a dictionary panel for quick lookups of definitions (found through Daring Fireball). I like right-click spell check/google/dictionarying and being able to click icons in the Task Switcher thing (Cmd+Tab). also, car repairs are expensive.

one of the things I think that was really glossed over at WWDC was Xcode 2.1. the really cool thing about it (besides the Universal Binaries thing) was the inclusion of WebObjects with the download - WO is now a free download with Xcode, and OS X Server now includes a deployment license for the product as well. (this makes deploying on platforms other than Mac OS X somewhat questionable, but, then again, I haven’t looked at it much.) thanks to will for the heads up on this one.

so, prompted by a (breif) discussion I had with a customer while I was extolling the wonders and benefits of .Mac, I gave a cursory glance at Plaxo. if memory serves, Plaxo made the rounds a few years ago when it was new. the idea is pretty simple - you install a plugin into your mail client (currently, Outlook or Outlook Express) and you use their servers to keep your contact data up-to-date. the twist is that your contacts are actually prompted to update it themselves. the idea is clever, since having your contacts manage the data about themselves minimizes errors. the problem is it only works with Outlook and Outlook Express, and only on Windows. (just like the problem with .Mac is that it only works on a Mac and with iCal and Address Book unless you get something like PocketMac.) there are some other problems with it too that may not actually be problems - problems like what format is the data stored in and how do I get at it beyond their web interface and Outlook (things I can’t check since I don’t have a running PC anymore).

so, what really needs to happen is that someone needs to implement a similar service that works with open standards. why not move data back and forth using vCal and vCard formats - pretty much everyone who matters supports them. bonus points if you make a plugin for Outlook (et. al), Entourage, iCal/Address Book/iSync, and Mozilla Thunderbird/Sunbird.

while we’re talking about calendaring, let me just say that iCal in the enterprise (or even in the I-need-to-coordinate-multiple-people-prise) sucks. iCal is great at what it does as long as you’re just scheduling yourself. once you get past that it’s less good - sure, there’s the capabilities (in iCal 2) to invite people (which involves sending everyone a nice vCal file - which in and of itself isn’t a bad idea) and to publish your calendar (which unfortunately is a read-only thing) but there’s not much else besides that. if I have a calendar published and someone else on my project needs to update it, they need to schedule it on their iCal and then send me an invite so I can put it on the “official” calendar. there should be a way to have a server (it can even be based on WebDAV) to do this - just have a folder that represents the calendar itself and then vCal files for each event. you’d still have the generate-event-invite-maintainer model for people who you don’t want to give calendar access to but everyone else should just be able to publish it to the server and everyone updates. speaking of I wonder how hard it is to write plugins for iCal..