Cost of IT Work?

[xopher]

Your teacher in HS was wrong then, too, and was likely teaching their students a habit they picked up during typewriter use. No one would have told them that it was grammatically correct to do so, unless they were born in the 1800s—and back then, it was only grammatically permissible.

If you do have that habit, it is indeed an important one to overcome.

All of my word processing/keyboarding classes were done on typewriters. :\

[Ratatapa]

In french the double space after the period is still valid. on handwritting or typing

[Mojo]

I bought the SSD hard drives with durability in mind, not speed. I've made SSD hard drives fail before. It took continuous writing of terrabytes/hour of data for months straight. The Dentrix application doesn't write that much data (the cause of SSD fatigue), but it does so in a way that the disk becomes very fragmented. His server HDD was so fragmented when I initially came in that it took an entire day to ghost. Mechanical failure from moving parts seemed like more of a concern to me. I tried to mitigate this by having a separate drive for the media files used by Dentrix image (the way this drive is set up, data will only be written to each block once or maybe twice.) Tape backup is not a feasible option. I was called in yesterday because they said their printers weren't working. They didn't turn on the computer that had the printers connected to it. Someone actually set them up with some onsite backup system. They ran it three times ever. In terms of the system drive for the server, very little information is rewritten each day. I would guess no more than 50 megabytes. Ultimately I don't see hard drive failure as probable. The application requirements haven't changed much since Dentrix 11, which has been ran mostly on the shitty Dell PCs since it's creation.

The backup client works like this. Every file in the backup set is hashed against the one stored on the server. If the hashes don't match, the new version is uploaded. The last 30 versions of every file are retained. The MSDE database for the image server is backed up twice daily, so he literally have two backups per day of that which will never be overwritten. I've been checking on it every day now to make sure it's been doing it's job. It's fine so far.

You guys have legitimate advice, but I think the system I have put in place will do just fine. Would you honestly think there is greater than a one in a million chance of a catastophic unrecoverable failure?

[Arkanna]

Actually, yes.

(I'm not being a smartass)

[Arkanna]

As to sound less argumentative and to continue contributing to the thread, I'm going to paint you a scenario, Mojo. Think it through ...

You get a call late one day, from the dentist. He's saying none of his workstations can connect to the application running on the server. You decide it's best to run in and take a look, thinking maybe someone just shut off the server or something stupid. You come to find out, the server won't boot. Since the best you can do is call upon a manufacturer warranty for a failed part, you diagnose the problem as quick as you can and discover the boot SSD took a crap. So now, unless you inventory spare parts at home (doubtful), you have to order a replacement SSD to get him up and running ASAP (worrying about the warranty claim after the fact).

But, it's late in the day and your order doesn't process until the next morning. Assuming the part ships ASAP and arrives the next day, that's already a full 24 hours your dentist has been unable to properly do business.

Now, you get the part in ASAP and replace it. Let's assume best-case scenario: the server is NOT a domain controller. You download 50GB of backup data, taking probably another couple of hours depending on the connection. This is now the majority of two days the dentist has been down.

In this world, most businesses won't tolerate that kind of bullshit. It will quickly turn into him figuring how much money you've cost him in business. If the scenario drags out longer, you could end up in court over it. I understand he's somewhat of a family friend and the likelihood of this even happening is low, but for fuck's sake dude, it can be PREVENTED in it's entirety by simply investing in a PROPER, HARDWARE RAID 5 controller and proper enterprise-level hard drives. It's a no-brainer, especially when you're doing this on a consultant basis. I guarantee, your dentist would be glad to pay more up-front for a proper server, instead of even risking a .1% chance of this scenario happening.

[hey]

You download 50GB of backup data

He said the dentrix data is 50 mb, and he already has 2 ssds in there. This sounds more like a few hours, worst case, to get back up and running. That's still way too long, considering a simple raid 1 would cost almost nothing, and remove any risk of downtime.

Realistically, unrecoverable failure is highly unlikely. However, recovery is going to be expensive, the way you have it set up. There's really no justifiable reason not to raid, especially when we're dealing with such little data. It'll cost all of what? $60 for a second ssd? If you really wanted to go all out, it'd be a few hundred for a good SLC ssd. And that would pretty effectively allow him to keep running if a drive does fail.

[Arkanna]

He said the dentrix data is 50 mb, and he already has 2 ssds in there. This sounds more like a few hours, worst case, to get back up and running. That's still way too long, considering a simple raid 1 would cost almost nothing, and remove any risk of downtime.

Realistically, unrecoverable failure is highly unlikely. However, recovery is going to be expensive, the way you have it set up. There's really no justifiable reason not to raid, especially when we're dealing with such little data. It'll cost all of what? $60 for a second ssd? If you really wanted to go all out, it'd be a few hundred for a good SLC ssd. And that would pretty effectively allow him to keep running if a drive does fail.

Your assessment is probably closer to true than my own scenario, but my point stands about RAID. His dentist may not need an uber server, but a simple hardware RAID 1 volume is absolutely a minimum.

[SephYuyX]

First rule of thumb: Never do business with people you know other than friends or family. Dentist, barber, mechanic, school nurse, etcetc, are a no-no.

You also cant get away with billing for hours of "learning". Some is allowed, but not excessive amounts; theres a difference btwn learning how the business/environment works vs learning how how to migrate from 2000 to 2008. In your case, saying it took you 70hrs when it could have been done in 10, and wanting to be paid for 70 is not going to fly. Learning on the job is the best way to do it, but you cant make the customer suffer for it, suck it up and take the hit. As for prices, I always charged $65 an hour, way under any other large or small store places.

Never become the person responsible for getting business back on their feet. Sign everything as is, no support, etc, unless you feel like getting dragged in to a shit fest. SLA agreements are a pain, and your ass is on the line if something happens, just avoid it. Say they can contact you when something needs to be fixed, but you cant guarantee how long it will take, etc.


Most of you posting are thinking on a enterprise scale, were talking about a dentist office here. Shit, ive set up small places with a scheduled task running a batch file that just copys relevent data to a 2nd HDD once a day. RAID makes things simple, but hell, if it's all just on a DB, run SQL Backup and call it done.

Many ways to skin a cat, but on small crap like this, it really doesnt matter what way to do it.

[hey]

Your assessment is probably closer to true than my own scenario, but my point stands about RAID. His dentist may not need an uber server, but a simple hardware RAID 1 volume is absolutely a minimum.

Right. It's so cheap, and will save so much time and money if anything does happen, there's really no reason not to.

Most of you posting are thinking on a enterprise scale, were talking about a dentist office here. Shit, ive set up small places with a scheduled task running a batch file that just copys relevent data to a 2nd HDD once a day. RAID makes things simple, but hell, if it's all just on a DB, run SQL Backup and call it done.

You don't need a large scale operation to benefit from local backups.

[Arkanna]

$1500 worth of hardware RAID controller/SAS 15k drives is hardly "thinking enterprise". Getting a server from a solid OEM that offers fast-response hardware coverage just makes good business sense. He'd be absolving himself of having to fuck with it, ever, and more importantly, if something DID happen, he wouldn't be the one on the hook to get the part sourced and replaced immediately.

You're right, there's many ways to skin a cat, but you don't do it with a potato peeler when a machete's available.

[Mojo]

I have the system drive ghosted on a disk and the Dentrix data (not the media files on Dentrix Image) is quite small. If the system drive did go out in the scenario you describe, I think I could have it up and running in a few hours, to be honest, but I do see your point. I was thinking more along the lines of whether or not it will be recoverable, not how much it would inconvenience him if something did happen. He plans on closing his practice in a few years though. I suppose that if he continues for more than a few years, I'll recommend he replace the server with a new one. Why use a machete when you can use a paring knife.

[Kohan]

Most of you posting are thinking on a enterprise scale, were talking about a dentist office here.

If I offered to back up an entire enterprise's network system with a single tape drive, I think I'd deservedly be told to go fuck myself.

[SephYuyX]

More of along the lines of thinking "fuck it, anything will work good enough" for this case. RAID is an easy and basic setup, but hell, even in this case it's probably overkill.

[Mojo]

This is what I have. Any suggestions?

Hardware Invoice
Labor Invoice