[NTLUG:Discuss] Data Storage recommendations wanted.

Thu Nov 16 00:04:02 CST 2006

On Wed, 2006-11-15 at 18:28 -0600, Robert Pearson wrote:
> On 11/15/06, Kipton Moravec <kip at kdream.com> wrote:
> > It is a garage setup to begin with.  He is just starting out on his own
> > after doing it for a lawyer in the evenings. Currently he is scanning
> > and putting the data on CD. Each of the lawyers clients is on a CD, so a
> > lot of space is wasted on many CDs. And they are filed in large CD
> > holders in alphabetical order. The current lawyer has about 500 to 700
> > CDs so far.
> >
> > I was under the impression tapes were very unreliable for long term
> > storage. The tape material becomes brittle over time and it needs to be
> > exercised (wound and rewound) every 6 months. This info may be dated,
> > but that is the way it used to be. (I am an old guy.)
> 
> I once thought CD/DVD had longer and better archive properties than tape.
> Not true. It says here:
> 
> Network World's Storage News Alert
> IBM expert warns of short life span for burned CDs, 01/10/06
> <http://www.networkworld.com/nlstoragealert16241>
> Opinions vary on how to preserve data on digital storage media,
> such as optical CDs and DVDs. Kurt Gerecke, a physicist and
> storage expert at IBM Deutschland, has his own view: If you want
> to avoid having to burn new CDs every two years, use magnetic
> 

I guess it is time for a really stupid question. Have you seen any data
on how long data stays on a disk if the disk is not plugged in?  I would
think it would be better than a tape since the media is sealed and has a
better quality surface to start with.

> The real issue is determining how to charge for this so the purchaser
> will pay for it and you can make money.
> Who controls the source information? Who owns it?
> Once the CD/DVD is burned does the source go back to the supplier or
> stay with the CD/DVD burner?

I think he was thinking of one copy to the customer and 1 copy to his
storage.  When I was at TI Germany, we had some very expensive parts put
in nitrogen storage. There were shelves with clear doors that sealed,
and they pumped small amounts of nitrogen into the sealed area to purge
all oxygen and moisture out of the storage to make the parts last
longer.  I wonder if that would work with CDs? It sounds like they have
to be cool and dark also.

> 
> > I am guessing the start will be on the order of 1 TB on line. And as the
> > business grows the storage will need to grow in increments. I don't see
> > a problem with a 10TB storage limit. If he needs more than 10TB get
> > another server.
> 
> A recommendation would be to make the Storage totally independent of
> any servers except the Storage server. And the Storage servers must be
> redundant to avoid losing access to the Information. I'm not talking
> about "99999" (5 nines) of availability or even "999" (3 nines). I'm
> talking about commodity priced hardware is cheap and fails more often
> than more costly hardware. Usually you can buy 3-4 commodity priced
> servers for the cost of one high quality. So buy at least two for each
> Storage node. If you are going to have direct-connected Storage. NAS
> begins to look pretty good, price wise, when you start looking at the
> cost of recovering 1-10 TB from tape. First you have to fix the
> Storage that failed, or server the Storage is connected to that
> failed, and then you get to spin tapes for hours and hours and hours.
> By the way it comes off the tape faster than it goes on. Keep that in
> mind for the Backup window.
> 

That is what I was thinking about. For a single datapoint, Fry's
Outpost.com is selling a Seagate 400GB ST3400633AS-RK Serial ATA
(SATA/300) Retail Hard Drive.. for $129. Three of those in a server and
you have 1.2 TB. 

That is about 571 CDs (If you packed them all the way to 700 MB) Or 615
CDs if you averaged 650 MB per CD.

Fry's also sells 100 HP CD for $25. They were the most expensive because
the article said the more expensive media could last up to 5 years as
opposed to 2 years for the cheaper media. So assume $0.25 per CD, that
means 571 CD's cost about $142.75 and 615 CDs cost $153.75, so the media
cost is about the same. It takes a lot more labor to burn 400 GB onto
571 CDs than onto a single hard drive. 

It would be cheaper to buy a hard drive load the data onto it, unplug
the hard drive and put it in an antistatic bag and place the hard drive
in storage.

> > Kip
> >
> > On Wed, 2006-11-15 at 15:48 -0600, ntlug at thorshammer.org wrote:
> > >       Define "business" :) Is it a garage type setup where you need the absolute cheapest solution held together with duct tape or is this something where you are just trying to keep a tight budget but you are expected to deliver to a customer?
> > >
> > >       Backups to CD/DVD is the cheapest solution but the least reliable for a business. Backup to tape is more reliable way but is more expensive.
> > >
> > >       As far as storage, you need an estimate of how much you need. Gigs, terabytes, 10s-100s of terabytes? You also need to think about RAID and the storage associated with that. SAN is the first thing that comes to mind but it's costly. There are also NAS and storage servers if you need smaller amounts of storage (less than 10T).
> > >
> > >       Basically there are pleaty of options for what you need but it's all based on budget.
> > >
> > > Robert Thompson
> > >
> > >
> > > On Wed, Nov 15, 2006 at 02:59:40PM -0600, Kipton Moravec wrote:
> > > > A friend of mine is starting a document scanning and storage business,
> > > > and asked me the best way to store the documents. The documents will be
> > > > scanned and stored on some type of non-volatile media, and will be
> > > > stored for so many days (30, 60 or 90) online.
> > > >
> > > > Right now he is backing up to CD.  I am recommending to upgrade to DVD.
> > > > Is that what you folks recommend?
> > > >
> > > > For the online data he does not need speed, he just needs tons of cheap
> > > > space. What is the cheapest way to store data on line? Just get a bunch
> > > > of SATA or IDE hard drives on multiple cheap servers?
> > > >
> > > > Kip
> 
> _______________________________________________
> http://www.ntlug.org/mailman/listinfo/discuss
> 
-- 
Kipton Moravec <kip at kdream.com>