Matt,
When it comes to the integrity and safeguarding of corporate data,
almost no price is too high. Luckily, cost effective means of backing
up gigabytes of volume, as well as ensuring continutity in normal
business operations, is well within the budget of every organization.
We use a three-pronged approach, which has served us VERY well over
the past five years:
1. Implement data redundancy at the server level, i.e. investing in a
hot-swappable RAID 1, RAID 10 or RAID 5 array using high quality SCSI
disks. This does a couple of things:
(a) Increases hard disk subsystem performance for all users (esp.
w/RAID 10)
(b) Provides online fault tolerance in case a disk dies, meaning that
we can replace a disk without bringing down the server at all
As a side story, two years ago I went on vacation to the Virgin
Islands. When we checked into the hotel, I had a message from the
office to call them. Turns out the night we left there was a bad
thunderstorm, which knocked out the power, which also caused a disk to
malfunction (it did not come back online). I knew it wasn't going to
cause downtime, so I had one of my cow orkers order a new drive, and
they installed it and it rebuilt the array while they could work. The
only real problem was the beeping RAID card, which was annoying and
easy to fix.
2. Invest in a high quality, high capacity tape drive. We run full
backups each and every night to a 200-40GB DAT drive; although DLT
drives have pretty much become a better option. The previous night's
backup is taken home with me, which provides full offsite warehousing
in case the building burns down. We also invested in a fireproof safe
for the tapes themselves.
We run full backups each night simply because it's much easier to
restore the files. Running differential or incremental backups saves
tape space, but it can become more problematic to keep track of all
the tapes.
3. Use multiple servers. We took an aging dual P-II 400 and made it a
backup Win2K server. It serves as a Domain Controller, DNS server,
print server, and performs some other functions of the "main" Domain
Controller. I added in some large, cheap IDE drives in a mirrored
configuration (using Win2K software RAID 1) and so I can run a
separate backup to that server every night as well.
This allows me to have an online backup for quick restores, which is
infinitely faster than retrieving it by tape. Being automatic and
always online, I don't have to really worry if, for some reason, I'm
not in and the tape doesn't get changed.
Having the second server be a Domain Controller also means that, from
an IT standpoint, my main Domain Controller's data and functionality
is backed up through Win2K replication automatically. This means that
if my primary Domain Controller machine (which is our main file
server) goes down, people can still log on, retrieve email, and even
work off of the backup data if required. I can then sync the data when
the main DC comes back online.
I recently had my file server doing hiccups due to an aging
motherboard/RAID controller issue, and having that backup server
available saved us at least three day's worth of work, since I was
rebooting the main DC many times trying to analyze the problem.
It's not about just backing up drawing files anymore; you have to
strategize on what else is a problem when your file server goes down.
Matt
mstachoni@comcast.net
mstachoni@bhhtait.com
On Fri, 29 Aug 2003 08:45:11 -0700, "Matt Vozzella"
wrote:
>Just curious, since the subject came up:
>How often does your office backup?
>How many different ways are the backups done?
>
>In our office we have:
>Server with mirrored drives - two copies constantly updated on main server-
>Alternate server with nightly backups
>Monthly CD's to be removed from office
>
>Our last major crash wasn't a big problem - cumbersome but all work was
>recoverable - except for the accounting side (We have since updated so THAT
>will not happen again).
>
>I'm just trying to keep up on alternate ways and ideas to set our office up.
>