Looking for ideas for Backups

wow, that is a lot further out than I would have imagined.

But then again, we do the full backup everynight, and I think there is only 5 tapes in the rotation, so that is about 70 usages per year.

How many tapes in your rotation?

FYI, I am concerned with being able to recover from a crash, if that changes the focus or not.

I did realize that we do NOT have a boot disk for the backup.

Need to analyze this from the ground up.

We currently have 485 LTO3 at around 600GB each, half in a library and half off-site. we also have a 10TB mirrored disk cache. And all of our data disks are replicated off-site.

Designed so I only have to be "tapebitch" once a week.



When you have actually BEEN blown up by terrorists then you take care to have a good disaster recovery system in place.

Is this your Domain Controllers, your file servers database server or what?

I approach disaster recovery different than data backups.

Disaster recovery is when a domain controller or database server dies and the entire network is unusable. Critical business data also falls into thei category. If that data is lost the business itself can not continue, that's included in Disaster recovery.

Data backups are the stuff that while inconvinent if lost and troublesome to recreate, are not necessary for the business to continue operations.This is usually user files and folders, Outlook contacts and calendars which might be the users responsibility or backed up to some media that is quick and easy to restore. Flash drives, external harddrives, or some other easily used rewritable media.


While system state backups are expedient, they might not be necessary in some instances. Domain Controllers that crash and you only have one of them, then you need basically a clone. If you have redundancy, then heck you can just install the O/S, patch it join in promote it and replicate.

The bottom line in planning is how long would it take to recover without undue financial loss to the company and without loss of data that can't be lost. Then a method is used to accomplish that.

Some like cloning, some like tape, some go to great lenghts and expense with tape libraries. Some use redundant systems.

you are depressing me.....

the tape backup is on the domain controller.

no backup of the terminal server, web server or email server.

Disaster recovery is non-existent.

Unfortunately, coming from a decade of *nix only servers, I don't know domain controllers and terminal servers. Spent a good deal of time reading about TS yesterday.

Their like web based applications. Everything is hosted on the Terminal Server except the GUI.

But back to backups. Terminal Servers are probably the worst after database servers to restore after a disaster. If it was me, I'd clone the the whole server if I could.

Web server... If it's 1 web site without a lot of fancy stuff, running on IIS, I'd just backup the site. The web server itself can be easily restored from media and patched up quickly, then all you have to do is put the site into a folder, tell IIS the path and your back in business.

Exchange works with Active Directory. So all user accounts and mailboxes are from Active Directory. Again, if users backup their stuff to pst files, you aren't going to lose much. The real problem is getting email boxes matched up to user accounts. In essence, if the Exchange takes a dive, and you didn't "clone" Exchange, you almost have to recreate everything from scratch on Exchange, eg set address space, go to each workstation, configure outlook again for each user, import user stuff from pst files, then your good to go. While not so bad with 9 users, anything more it can be a real time waster.

If I remember correctly, you only have 9 users. So recreating stuff isn't that much work if the Domain Controller takes a dive unless you have some complex GPO and share configurations. Even then you can export GPO configurations and save em. Share folder stuff you document so if you have to recreate it, you got the template in the documentation. The files in the shared folders are always backed up. User files that users say they can't live without are always backed up.

So, how I plan out disaster recovery is how I would go about recreating the entire network as if I'm creating the network from scratch. Those areas that require a lot of configuration, or is critical business data that isn't standard stuff, like databases and user critical files, I backup. Those I don't and use default settings, I don't backup. All this is documented in a step by step instruction.

With somewhat large small businesses, 25 users or more, redundancy is the life saver. Everything works while your trying to fix what broke and nothing is lost.

added note: just read GGs stuff. I assumed that you would understand that business critical data, data the business can not afford to lose, includes all the legally required records. As GG also points out, users or management have to tell you what their needs are so that you can meet them.If users need to access archived stuff 25 years in the past, well you gotta plan for it. some archived material doesn't need instant access, some does. Users store lots of crap they never use but say they must have. Saving data can turn into a ponzi. Data on top of data on top of data with this data growing expotentially right along with the cost. It can and does get out of hand real fast.

if you just back up the data, 5/10 years later, will you have the software available to read it?

Thanks!

After I get a few basics down on TS, my big thing will be to do the domain server. It is doing VERY little, really.

More on THAT later.....

just robocopy straight to external hard drives

Well, I assume you have a physical box. There may be other products but this product can take an image of a server while it's running:
http://www.ultrabac.com/

We use that to take an image of all our physical Windows servers. VCB takes care of all the virtual ones.

We're only running Symantec Backup Exec 10d but it suits our needs well. We have a tape library so our rotation may vary a bit from how you do yours.
We leave our daily tapes in and rotate the weekly/monthly/yearly...this is how ours works.

Mon, Tue, Wed, Thur are daily incremental
Fri daily Full
Sat is a Full weekly
Sun is Monthly incremental.
We take the weekly tape to the bank (safe deposit box) on Mondays.
The last Sun of the month does a Monthly ....the monthly goes to the safe deposit box....Last Sun of the year is a Yearly.

We have 4 weekly tapes in constant rotation, 12 Monthly tapes in rotation, and each year gets its own tape.

As you can see, we have a lot of tapes in use but they want to make sure we can go back to almost any time in history if need be....we're dealing with tax files etc.

do you have it set so that it does each increment after each other, or overwrite the previous?

Bad thing is, I am not even sure how much data needs to be backup up yet! oh my....

That way we can restore to any given day more easily.

The amount of data is a whole other issue....trying to convince people we really don't want to backup copies of "their" backup of "their" backups...it can be quite a challenge to get them to trust the fact that you can restore it for them.

is essentially the same thing as 11d.

We run both here and honestly, I like 11d better. The problem I have with 12 is anytime something has to do with SQL. The thing is so picky. The backups "fail" but still show that data is being backed up. I never have that problem with the 11d machines.

For your original question we run a series of 14 tapes on the 12 machine and 7 on 11d machine. Tapes get retired after we take note that a backup for that particular tape fails more than once. On a failure we try to do a manual backup on the tape, if that fails we grab a new tape from the supply, relabel it and go from there.

If you are using tapes right up to failure, do you get concerned about getting false positives as to your backups?

Thanks for the info on v12. As this is a SQL server2005, I will stay at v11d. Will read up on the service packs though. I am on SP3, and SP4 is out.

Usually once a week someone deletes a file they shouldn't have or we need to make sure certain files get backed up. That serves as our verify.

And if you do, expect to use more tapes, because the encrypted data doesn't compress as well.

"A tape should be retired after two to six months of use to prevent media failure."

months of use, so if is used one day a week, are they talking calendar months or actual usage days?

If the time frame is as vague as 2-6 months, what difference does counting/not weekends???. Since few would be backing up on the weekends They would more than likely be inclusive. Don't you think? Six months seems too short of a life span to mee.Commvault, and Arcserve expire tapes for you. They keep track of usage and notify you when it's ready for the"Old Tapes Home". CFheck and see if symantec has the same feature.

Sure would be nice if I had the books/disks that came with the symantec! grrrr.

Well, help file, here I come!

Thanks for the tip!

jd

If you are looking for the product installer, I would use Symantec FileConnect:

https://fileconnect.symantec.com/licenselogin.jsp?localeStr=en_US

You will need the product's serial number though.

If your tape drive has a barcode reader, each tape can be cataloged according to a barcode. BackupExec (at least 12d) will record all sorts of information about each tape, including how many hours the tape has been in use and how much total data has been written to it and read from it over its lifetime.

If it encounters hardware based errors, it will also record that.

I don't think this has bar code reader?

You DO know how to use a spreadsheet program, don't you?

I think you're in an RTFM situation.



Now, if you could take a thumb OFF for this answer, you probably should.

as I clearly said before, I don't HAVE a manual to F'en read!

No software and no documentation. Things would be a lot easier if I did have it.

And just to make you happy, I gave you a thumb just so I could take it away!

I think you said you had 11d.

Good old Admin Guide:
http://seer.entsupport.symantec.com/docs/285233.htm

Maybe I will pass on printing it off.....


Thanks for the link!

Sure thing mate.

A lot of companies label the manuals to Network Admin applications "Server Admin Guide", "Admin Guide" etc. So typing the product name plus some variation of those will often yield results.

Crafting google searches is an art.

the INDEX is 45 pages!!!!!

I was going to print just THAT off to make it easier to look things up.... wow.....

Anyone feel like reading this and telling me what it says?

NTBACKUP is better, cheaper, and faster

BE stores media statistics... starting from either when the tape was formatted, or when it was imported.

On the media tab, right-click on a tape and select properties. There should be a statistics tab.

What kind of rotation of drives do you use?

Do they have software that can boot off a boot disk for a baremetal restore?

http://www.backupexecfaq.com/articles/using/performing-a-local-disaster-recovery-with-11d.html

the os is on a mirror the data on a raid 5

the os drive doesn't change much if any

most rotating three drives, one offsite at all times

occasionally i image the servers with acronis workstation w/universal restore

so yes i can actually restore bare metal even to new hardware and have done so several times

No one knows legally what is required. They have paper copies of transactions, and that was traditionally the only thing that was considered.

A call to our CPA is going out, and we hope to get the legal question answered.

What is the life expectency of the drives vs tapes?

If you got a longer life and more capacity, I could see that being a real benefit for people with large data backup needs, over a few year span, but would probably take a few years to pay for its self.

And yes, I have always been a fan of full backups. If they run at night, why would I care if it takes one hour vs 20 minutes to run?

We do get quite a few support calls on Backup problems, including:
"Forgot to put tape in drive"
"Tape out of Capacity"
"Bad Media"

And thats what lead us to look at alternatives. We first had the GoVault, similar technology, cartridges from 40 to 160GB. They really are just platters from a HD encased in a cartridge that go into a carrier with the HD spindle/motor/controller.

The Go Vaults had a 1 year warranty, not sure about RDX.

James

looking around the server trying to find out where all the data is. The databases aren't that big, but the drive was showing 300 GIGS used???

There they were, almost 300 One gig bkf files at the root directory, all dated yesterday around noon when I was having failed backups and running test backups on tapes to see if I have any that still work (I ordered a new box that should be in today/tomorrow).

Looking at the BE settings, I can't see ANYTHING that would tell it to make these?

Can I delete them?

On the BE console, go to the Devices tab, expand Device Pools -> All Devices. That's where they are defined. Again, I don't know what Quickstart edition does or doesn't have.

When I went into the job, it was set to backup to all devices, which is the tape AND the backup to folder.

I changed that to only point at the tape, and disabled the folder.

So, it should be safe to kill the 270+ files filling up my hardrive? The backup files are taking up more space than I have data!

Side note, WITH compression, the tapes only hold about 70 gigs.

Backup to a folder isn't a bad idea. I am not going to advise you to delete it simply because I am not right there looking at it. It would be foolish of me to suggest deleting a backup without really knowing what it is. You'll have to make that decision.

The folder backup isn't a bad idea. If it's exactly the same as the tape backup it's really a convenience issue. Do you get users asking you to restore files a lot? Can you see how it would be a cumbersome task to do those kind of backups from tape? Especially considering you already told me you don't have a bar-code reader and no way to automatically catalog them.

to make good and sure I have a valid backup.

The problem I see is two things.

1) dumping the backups at the root instead of in a directory

2) not having that directory excluded from the next backup.

I am thinking my backup failures are from exceeding the capacity of the tapes because of these backup files.

Will then move the files, set an exception, and see what happens. after a week, I will kill them and start doing a NEW rotating and intentional disk backup as well as the tape backup.

Thanks for all the help.

jd

Backup to Harddrive instead of tapes? Are you serious?

First of all, I sure hope you are getting solid state drives. Something about moving platter based hard drives around and tossing them in bank vaults doesn't seem right to me.

Sure, it may backup faster. But is it really all about speed? Don't get me wrong, speed is nice, but isn't it really about having a copy of company data in case of emergencies? I wouldn't want to be the guy whose hard-drive backup died.

If you are getting solid state drives, well I hope you're loaded. A 128 GB SSD costs around 400-500 dollars. All that for a measly 128 GB. And by the way, you're going to need more than one.

Let's assume all your data somehow fits in 128 GB, you're going to need 5 for differentials/incremental, 1 for the Full weekly. And how long do you keep the weekly tapes? Let's say you keep them around for a month. Well that's 4 more. And what about month ends? 12 months in a year, that's 12 more drives. 21 tapes in total times 400 dollars, that's nearly 5 grand. And that's if all the company data fits in 128 GB.

Tapes on the otherhand.

Let's say you have LTO3 system. A single loader LTO3 tape drive costs around 1,500 dollars. LTO3 tapes hold 400/800 (400 GB, 800 compressed... whole lot more likely you're going to fit all your company's data on one of those.

LTO3 tapes are about 30 dollars each. We established you're going to need about 21 of them. 21 times 30 dollars is 630 dollars. So you're looking at somewhere around 2,000 - 2,200 dollars.

Oh, and if your company's data grows and say the full weekly and month ends start to take two tapes, you'll need to buy more tapes, and at 30 dollar a piece, that's manageable. But I'm sure whatever big C you need to report to is going to love to hear that you need to spend money on a handful of new HDs costing 400 dollars each.

If you're not actually getting solid state and cheap platter drives... we'll that's just foolish. I don't like to play Russian roulette with my job.

Tapes are around for a reason. Doesn't matter how fast the backup is if it failed.

I do not back up once to one media, but twice to two separate sets of media and keep one offsite. I also test restore my servers and, believe it or not, I have several accounts wherein I can restore a server in 10 minutes from total crash to working fine.

Also I know a little tiny bit about data restoration. Before you get on your high horse here.

101st floor, South Tower, World Trade Center.
System admin for Aon Consulting.
So maybe I kinda know something here.

You didn't mention HDD as a secondary backup. I was under the impression that you were suggesting platter based hard drives as primary backup... and to be honest I don't think it's a good idea.

Having that as a secondary option isn't a bad idea.

SSD drives are bound to come down in price. When this happens I think it would be a lot more practical to use them for backups like tapes are being used now.