PDA

View Full Version : need a massive storage/backup solution for growing branding firm




seenew
Jun 6, 2012, 10:33 AM
Hey, so.. I work as a photographer and videographer for a small (but quickly growing) branding firm in Manhattan.

We're in desperate need of some sort of file storage/backup solution. Since we're a decently small company and only just got into media production in the last year, this is something new that no one here has dealt with before.

Most employees either backup to portable drives or don't backup at all (we have no actual IT department to lecture them). They're probably fine using portable drives to back stuff up, but our art department has grown to two videographers, two graphic designers, and one web developer, and we're producing tons of data, especially for video. I'm talking like 25-50 gigs per shoot, on average, and 2-3 shoots per week usually. So just downloading the cards takes up enough space without mentioning duplicating it (I've got a stack of six 1-2TB hard drives sitting beside my iMac, which already has 2TB of internal space filled).

And since we don't have an IT person and no one in the office is really that tech-savvy, it kind of falls to me, since my team has the most need for it.

What we need is some sort of massive storage solution-- somewhere I can dump my cards after a shoot, organize them (renaming, keywording, filing them in some sort of structure) and also back them up. Besides an onsite backup, it'd probably be good to have an online backup system, too.
Whatever drive the files are stored on needs to have a fast enough interface that I can either edit video directly from it or download a dozen gigs from it quickly. Also, if it could be accessed from an offsite machine (so I can work from home), that'd be great.

All of us in the art department have new iMacs with Thunderbolt capability, as well as the standards-- USB 2, Firewire 800, and ethernet.

I think we'd need to start with something like 20TB at least, and it'd be great to be expandable. Our company is growing fast, so we'll be producing more and more media at a faster rate.

The server would only need to be accessible to the art department, but if there is a way to have it accessible to the rest of the 40 or so employees, that would be helpful, too.



Sooooooo... I can only describe our needs, but I have no idea what we should get or if there is a company or someone we can call to come and set something up for us.. Or what I should tell such a company.

Money is sort of not an object. Meaning there is money to do this, but as I said, I am one of only a few here who appreciate the need for this and how much it could cost initially. I'm definitely expecting it to be at least a few to several thousand dollars. Hopefully 10k or less, though..

Any ideas? Thanks!



aarond12
Jun 6, 2012, 01:16 PM
Ever since the demise of the xRaid product, I would have to say you need to go to Dell or HP or a similar company for large storage solutions.

http://www.dell.com/storage

Remember to think about off-site backups (e.g., rotating tape backups that are picked up on a daily basis by a service). Natural or religious disasters must be accounted for. Keeping your data off-site is the only way to be really sure that you're protected.

blueroom
Jun 6, 2012, 01:20 PM
Synology makes an entire line of NAS devices. All excellent, reliable and easy to setup and use. They support Mac, Windows & Linux clients.

JKColo22
Jun 6, 2012, 03:29 PM
With the amount of data you are looking to back up and the amount of money you are willing to spend, you need to bring in a pro for this job. I would call a local Apple authorized reseller and ask them if there is someone they could recommend. You need to find a company that is familiar with working with macs and can provide client-side support in addition to setting up a backup platform.

mabaty
Jun 22, 2012, 12:39 AM
If we were to consult this gig, I would recommend something like a QNAP TS-1279U-RP strictly as a media server for the art department. Then, I would look into incorporating BOX.com (Enterprise/business) for whole company file sharing. We are moving a 200 employee digital marketing company over to BOX now, and it is fantastic. Of course media guys still need local data storage because of the amount of raw data they move, that's where the QNAP comes in. It will support 10GbE connections for you, and 1GbE for others just to access the data. I would consider keeping the art dept. media separate from account managers and other administrative files, although it doesn't have to be that way, as the QNAP supports sub folder permissions.

hchung
Jun 22, 2012, 01:47 PM
That QNAP unit looks promising. I haven't used one, but it's been mentioned by other people in the past too.

Despite the "money is no object", to get something that'll really blow your mind will probably exhaust your funds. If you want to see what real speed for shared storage looks like, consider the following:

1) http://www.oracle.com/us/products/servers-storage/storage/nas/zfs7120/overview/index.html

This gets you storage in hard drives, backed by flash, and checksumming/error correction as well as redundancy. You'd also need a 10gigE card to get the full bandwidth if you want more than a couple of macs to get the full speed.

2) Cisco Catalyst 4948 (48 gigE ports to 2 10gigE ports on a switch)
This gets you the ability to allow multiple macs to communicate with said server at the imac's full speed. (limited by gigE)

USB2 is out of the option because it's slow. Thunderbolt is out of the option unless you're using it for local storage or you're using to get a 2nd gigE port.
Firewire 800 is only good for local storage.
The only way to get shared network storage to your iMac is gigE, but you're limited to about 75MB/sec per iMac.

From there on out, your next bottleneck is how fast the server can push and pull to its drives. That's what the ZFS rack is for. It's hard for me to estimate what kind of speeds you'll get, but you should be able to max out the bandwidth to at least 4 iMacs simultaneously. Not good enough to video edit in my opinion, but plenty fast to copy stuff around as needed.

(No I don't do this for living, but I did build a dirt cheap consumer version of this for my own personal storage needs. It's awesome. But since you have real clients for this, you shouldn't be cobbling consumer hardware for your storage needs.)

belvdr
Jun 22, 2012, 02:12 PM
That QNAP unit looks promising. I haven't used one, but it's been mentioned by other people in the past too.

Despite the "money is no object", to get something that'll really blow your mind will probably exhaust your funds. If you want to see what real speed for shared storage looks like, consider the following:

1) http://www.oracle.com/us/products/servers-storage/storage/nas/zfs7120/overview/index.html

Not sure I'd go with Oracle hardware these days. It's a toss up as to what's coming next. Plus, everything they touch is generally broken.

2) Cisco Catalyst 4948 (48 gigE ports to 2 10gigE ports on a switch)
This gets you the ability to allow multiple macs to communicate with said server at the imac's full speed. (limited by gigE)

Might want to choose a current, non-obsolete switch, like the 3750-X. ;) I'm assuming that the people aren't gathered around that one switch, so you might need more than one. In that case, you could use a smaller switch for the clients, so long as it's not oversubscribed heavily.

hchung
Jun 24, 2012, 02:12 PM
Not sure I'd go with Oracle hardware these days. It's a toss up as to what's coming next. Plus, everything they touch is generally broken.



Might want to choose a current, non-obsolete switch, like the 3750-X. ;) I'm assuming that the people aren't gathered around that one switch, so you might need more than one. In that case, you could use a smaller switch for the clients, so long as it's not oversubscribed heavily.

Like I said, I don't do this for a day job. :) But yeah, the 3750-X looks good.

As for the Oracle gear, I'd recommend it as it's basically conventional server hardware already QA'd for Solaris. It shouldn't really matter what's coming up next, as you got what you asked for: ZFS, L2ARC, a couple 10GigE ports, and a bunch of drives. The software is what makes it unique, and they're pushing updates. If it just happens that it's broken when you bought it, they're obliged to fix it or you return it.

My experience is even on cheap consumer hardware (CoreDuo, 6 spindles, and a USB flash drive as a ZIL), I can provide myself with a safe place to store my data and be aware of upcoming faults, while being able to max out a GigE connection. If Oracle hardware makes you uneasy, there's always the option of getting say HP server hardware and trying to support Solaris on it yourself, but that feels like taking on too much responsibility that the vendor was supposed to handle.

guy999
Jun 25, 2012, 12:01 PM
I have a synology 2411+. It can hold up to 12 4 TB drives, using Raid 5, you still ought to have something like 40 TB of storage available on the network. It's pretty easy to get working and you can actually use synology hybrid raid and add drives along the way so you don't have to immediately spend the cash for 12 4tb drives.

you can also add a satellite 12 bay unit that will connect over a special link that will give you 24 4 tb drives, that ought to hold you for awhile.

phr0ze
Jun 25, 2012, 12:03 PM
I second the QNap hardware. Its been great for my all Mac environment.

I have 10TB in mine, but its a smaller one for home use.

TheCoupe
Jun 25, 2012, 06:08 PM
Hi, why don't you just build your own system... which is pretty much what I've done for a customer, although there's is for a backup to their existing raid.

I'm quoting this kit as it's what I've personally used and know it works:
Base spec Mac Pro
LSI 3ware 9750-8e Raid Card
LSIiBBU07 (LSI00161) Battery Backup Unit for 9750 card
Norco 24 Bay Hotswap Chasis
1 x SFF-8088 to SFF-8088 Cable
6 x SFF-8087 to SFF-8087 Cables
1 x Chenbro UEK23601 (Card and Daughterboard)
750w PSU (preferable to get a redundant PSU)
2Tb Hitachi 7k3000 drives, or equivelent

You will need a few connector cables on the power depending on your PSU, but this system takes about 2hrs to build, and is very very fast. I recommend multiple volumes... I use Raid 6 for main data, and Raid 5 for Backup, and have 3 x 8 Bay volumes per chassis.

Also, if you use an SSD as your server boot drive, it will scream :)
If you need pics, PM me your email.

Cheers

Justin

Alameda
Jun 28, 2012, 12:42 AM
Let me make a parallel suggestion: You don't just need storage hardware, you also need a strategy for managing your storage and workflow in your office. With so much data, if you don't come up with a consistent way to label it, you're going to have a massive mess on your hands.

One way would be to use Automator or AppleScript to create folder actions for dumping the cards into storage. On the file copy action, the scripts could name the files according to your naming standards, put them into named directories according to your rules, and you can perform other actions to ensure that everything is catalogued correctly.

mabaty
Jun 30, 2012, 01:12 PM
Hi, why don't you just build your own system... which is pretty much what I've done for a customer, although there's is for a backup to their existing raid.

I'm quoting this kit as it's what I've personally used and know it works:
Base spec Mac Pro
LSI 3ware 9750-8e Raid Card
LSIiBBU07 (LSI00161) Battery Backup Unit for 9750 card
Norco 24 Bay Hotswap Chasis
1 x SFF-8088 to SFF-8088 Cable
6 x SFF-8087 to SFF-8087 Cables
1 x Chenbro UEK23601 (Card and Daughterboard)
750w PSU (preferable to get a redundant PSU)
2Tb Hitachi 7k3000 drives, or equivelent

You will need a few connector cables on the power depending on your PSU, but this system takes about 2hrs to build, and is very very fast. I recommend multiple volumes... I use Raid 6 for main data, and Raid 5 for Backup, and have 3 x 8 Bay volumes per chassis.

Also, if you use an SSD as your server boot drive, it will scream :)
If you need pics, PM me your email.

Cheers

Justin


If you buy from a manufacturer, and your a business you get the benefit of the support. For example, one of my clients Proliant storage server drives failed in their San Diego office, I contacted HP had a new drive shipped there in 2 hours and replaced by end of day, the server is 3 years old. Build it your self and good luck getting repairs that quick.

TheCoupe
Jul 1, 2012, 07:35 PM
If you buy from a manufacturer, and your a business you get the benefit of the support. For example, one of my clients Proliant storage server drives failed in their San Diego office, I contacted HP had a new drive shipped there in 2 hours and replaced by end of day, the server is 3 years old. Build it your self and good luck getting repairs that quick.

With the amount you save, you could build 2 or 3 of these systems, set them up to mirror and keep a few drives / parts spare for less than what it normally costs to buy one system with support.

I'm in the UK and with using a standard PSU rather than redundant and powering the unit from a UPS, I can build one of these systems for about 1,000 + drives, and that includes the Raid Card.

Buying this in the US is cheaper again.

The other alternative, is to purchase an OFFICIAL raid system by some company, then use this system for your backups... but if it was me... I'd sooner build multiple systems and save cash than go buy one that could possibly have parts sent out if a failure occurred.

For reference, this build all came about after discovering Backblaze:
http://www.backblaze.com/
http://blog.backblaze.com/2009/09/01/petabytes-on-a-budget-how-to-build-cheap-cloud-storage/

and reading:
http://www.servethehome.com/sas-expanders-diy-cheap-low-cost-jbod-enclosures-raid/

and for that matter, what's to stop you building your own system and using Backblaze as an off-site Backup solution?

Hope this helps

DotComBoyUK
Jul 3, 2012, 04:01 AM
So a number of people have already mentioned a company I work for and the product I may recommend based on others doing the same thing (So I will declare I work for Oracle) with that in mind I would consider the below when looking for a solution (I won't recommend product due to my potential bias).

Life and access of the data you are storing is key. Powering Disk Drives gets expensive especially if you are doing it in two places. Along with this your access of the data will eventual result in the majority of the data not being accessed after 180 days. Data on disks slowly corrodes unless you have something that can correct bit flips. Some filesystems do this themselves (ZFS) but others may require a content management system that provides checksum and reporting as 1 bit flip can render a photo either unreadable or partially incomplete. (You just need to look at my Aperture Library with photos from 2001 in to see that effect).

Generally when I look at Data Protection I tend to follow a simple rule of 3 for any data I create:
3 copies of the data of which 2 are on different media (Filesystem or Physical media) and with 1 offsite

With the two statements above I would personally look at introducing some tape storage which has longer shelf life without corruption whilst reducing the total amount of data you are having to store on disk. At the same time a central disk platform will make things easier for you but don't think that RAID in a central array is the answer to all of your problems ideally you need to look for checksums on the data and preferably some form of snapshots so you can go back in time and repair, undelete files as well (Same type of functionality as time machine).

An alternative is to continue in a disk only manner but ensure that you have 2 filesystem types HFS and FAT as an example. That way a filesystem corruption that may occur on 1 data set will not corrupt the other.

mrichmon
Jul 3, 2012, 08:18 PM
An important question the OP needs to answer for himself is "how much of the data needs to be online at a given point in time?"

The 20TB of initial storage isn't all that big and is easily achievable in various configurations as suggested by other posters. The trick is expansion. If the OP really needs to expand then NetApp, EMC, and other storage hardware vendors have products that can be configured in the 20TB scale and will grow from there.

However, in my experience most videographic and photographic groups are better served by a modest "near-line" data store with movement of data files out of the near-line store to an archival store when a project completes. (Or 3 months after a project completes, pick your timeframe.)

Products and technology suitable for an archival data store are very different than those suitable for a near-line store. Some important differences are: a) unchecked/unused data rots over time when stored on general purpose storage systems (archival systems mitigate this issues), b) general purpose storage is expensive GB for GB when compared to offline archival storage (higher cost of online storage, wear and tear on online systems, power costs).

martinX
Jul 4, 2012, 04:39 AM
http://www.youtube.com/watch?v=Y-6EQo6it7Y

lannisters4life
Jul 4, 2012, 04:46 AM
Natural or religious disasters must be accounted for.
ROFL. Does Tomkat's breakup count as one of these?

rjc101
Jul 4, 2012, 12:22 PM
If you are really thinking of 20Tb for starters, and need to keep the information on-line, then I'd be looking at a NetApp solution. These can easily be configured for very high availability and are fairly easy to expand. Cost is the downside, but you get what you pay for.

There are a number of off-site backup services in the UK which offer, via SnapMirror, to keep an off-site copy of your data which some clients find handy rather than trying to back everything up themselves.

will waters
Jul 4, 2012, 01:54 PM
Get yourself one of these to start:

http://www.comms-express.com/products/24tb-netgear-readynas-4200-12x2tb/

Along with at least a gigabit switch: Possibly one of these as well (Unless you want to go with a managed switch which may be better)

http://www.comms-express.com/products/netgear-gs724ts/

Hope this helps

Will

ulzeraj
Jul 7, 2012, 12:36 PM
I can't recommend enough a ZFS based solution. Explendid volume management, integrity checking, snapshots, clones, can cache writing and read operations to a SSD device (almost SSD speed for spinning disks!), easy to add and replace disks and the list goes on.

There is a good number of flavors depending on how much skill and money you have available. From pure Solaris or FreeBSD to NexentaStor or FreeNAS.

TimTheEnchanter
Jul 11, 2012, 11:39 AM
We recently upgraded our archive storage and backup system.
- Staff of 18, creative group is 8 workstations.
- Each computer has extHD w/Time Machine.
- Projects are archived from workstations when complete to the server.
- Server is new MacMini Server with a 12TB Pegasus TB RAID (http://store.apple.com/us/product/H5187VC/A/Promise_Pegasus_R6_RAID_System?n=raid) being our archive server.
- Drobo Pro for local backup of server RAID.
- CrashPlan Pro online backup of server, to be expanded to workstations soon.

We already had the Drobo Pro as a prior BU, replacing expensive and unreliable tape BU we had for years. The MacMini and 12TB RAID cost us about $3000 after shopping around for a better price on the RAID. They have smaller configurations, but for the money it gave us plenty of headroom to grow (as file sizes keep increasing). Was easy to set-up and transfer existing data. CrashPlan Pro costs us $7.50 per month to backup now 4TB worth of crucial data, peace of mind for emergencies.

So far it's been serving us well, no issues other than on nagging network copy issue where random file and folder modification dates change upon copying over the network. Otherwise, very fast and feel we're covered much better than before.

sfxguy
Jul 11, 2012, 06:02 PM
We recently upgraded our archive storage and backup system.
- Staff of 18, creative group is 8 workstations.
- Each computer has extHD w/Time Machine.
- Projects are archived from workstations when complete to the server.
- Server is new MacMini Server with a 12TB Pegasus TB RAID (http://store.apple.com/us/product/H5187VC/A/Promise_Pegasus_R6_RAID_System?n=raid) being our archive server.
- Drobo Pro for local backup of server RAID.
- CrashPlan Pro online backup of server, to be expanded to workstations soon.

We already had the Drobo Pro as a prior BU, replacing expensive and unreliable tape BU we had for years. The MacMini and 12TB RAID cost us about $3000 after shopping around for a better price on the RAID. They have smaller configurations, but for the money it gave us plenty of headroom to grow (as file sizes keep increasing). Was easy to set-up and transfer existing data. CrashPlan Pro costs us $7.50 per month to backup now 4TB worth of crucial data, peace of mind for emergencies.

So far it's been serving us well, no issues other than on nagging network copy issue where random file and folder modification dates change upon copying over the network. Otherwise, very fast and feel we're covered much better than before.

I'm looking at setting up a very similar thing with 9 staff, 6 of them doing film composting.

How are you finding your artists speed at pulling large amounts of data off of the Pegasus? Is it slowing them down at all?
I was thinking of the new Drobo 5D when it comes out in a few weeks as opposed to the Pegasus.

How are you finding the crash-plan service? How long did it take to get 4 TB's of data onto their servers?

We could potentially be wanting 500gigs of data updated daily so I'm wondering if that is even reasonable to think of with a service like crash-plan.

Thanks!

TimTheEnchanter
Jul 11, 2012, 08:45 PM
I'm looking at setting up a very similar thing with 9 staff, 6 of them doing film composting.

How are you finding your artists speed at pulling large amounts of data off of the Pegasus? Is it slowing them down at all?
I was thinking of the new Drobo 5D when it comes out in a few weeks as opposed to the Pegasus.

How are you finding the crash-plan service? How long did it take to get 4 TB's of data onto their servers?

We could potentially be wanting 500gigs of data updated daily so I'm wondering if that is even reasonable to think of with a service like crash-plan.

Thanks!

Performance is great. I never notice bottlenecking or lag, no complaints from the crew. I've been a big fan of Drobo (have one at home... photographer) and haven't had any issues with it other than being slow (due to the way Drobos write redundancy at once). They're simple, self-monitoring and easy to maintain. That said, I am growing more concern of the day it does fail, as all RAIDs do, because of their proprietary closed system. I think it will be much harder to recover a failed Drobo RAID than a basic RAID enclosure. But the Drobo is a backup, given the Pegasus is the primary RAID, I'm confident that the data there between the two in a hardware failure. In the case of major failure or catastrophic event, I have CrashPlan.

I'm a huge fan of CrashPlan. I found them to be the best all around and faster in multiple tests I ran before going with them. I also like the flexibility their software gives me to backup in other means. As for how long, our initial backup of critical data, about 2.5TB, took about 1.5 to 2 weeks on basic business-class cable, but that was with some interruptions. First and foremost, I completed a full local backup and took it offsite, then held-off on adding new content until CrashPlan was complete. They do have a seeding option but the boss didn't want to pop for it and I felt ok with having our ars covered and the estimated time to complete.

Feeding 500GB daily is a lot to push to any cloud service, but it's a constant connection, between overnights and weekends it might keep pace given a higher-class business connection (upstream).

Ap0ks
Jul 12, 2012, 02:24 PM
Get yourself one of these to start:

http://www.comms-express.com/products/24tb-netgear-readynas-4200-12x2tb/

Along with at least a gigabit switch: Possibly one of these as well (Unless you want to go with a managed switch which may be better)

http://www.comms-express.com/products/netgear-gs724ts/

Hope this helps

WillWe have a ReadyNAS 3200 and although it does the job for our non-critical stuff, I would say it needs more work before I recommend it to others (the admin interface is atrocious) ;)

NorCalLights
Jul 12, 2012, 03:36 PM
I use Backblaze for my offsite backup. Their support was fantastic the one time I needed to contact them (on a Sunday of a holiday weekend, they emailed me back posthaste). Other than that, it's unlimited, cheap, and it just works.

www.backblaze.com

They seem like they're huge storage nerds... they've even open-sourced their massive storage hardware and their design is being used by universities and even other large storage companies.

trevorlc
Jul 14, 2012, 02:55 AM
I think it will be much harder to recover a failed Drobo RAID than a basic RAID enclosure. But the Drobo is a backup, given the Pegasus is the primary RAID, I'm confident that the data there between the two in a hardware failure. In the case of major failure or catastrophic event, I have CrashPlan.


Failed Drobo's can be pretty messy based on some posts I've seen, however my original usb2.0 only drobo failed about 2 months ago, I was able to purchase a new one and put the disk pack in and it was like nothing ever happened. So it can be rather painless too. As long as your willing to accept that you have to buy another drobo.

I did heavily consider buying a drobo get my data off and return it after I was able to transfer the data to a standard raid. After researching raid options I ended up liking the drobo the best, the pure fact that as it gets filled up I can pop a new drive into it is awesome.

TimTheEnchanter
Jul 15, 2012, 03:11 PM
Failed Drobo's can be pretty messy based on some posts I've seen, however my original usb2.0 only drobo failed about 2 months ago, I was able to purchase a new one and put the disk pack in and it was like nothing ever happened. So it can be rather painless too. As long as your willing to accept that you have to buy another drobo.

I did heavily consider buying a drobo get my data off and return it after I was able to transfer the data to a standard raid. After researching raid options I ended up liking the drobo the best, the pure fact that as it gets filled up I can pop a new drive into it is awesome.

Your Drobo recovery story is reassuring. So far, my experience has been great and I've had no failure...knock on wood. It's those stories I've heard that has me concerned, but your's gives me a little relief. Thanks for sharing that.

ricosuave
Jul 18, 2012, 12:06 AM
I've searched the web but have not read all the posts in this thread, sorry.

For those that have the XRAID, does it allow to pop in drives already formatted to Mac OS Extended and use them like a large external drive enclosure? Or do you have to format them in some sort of RAID configuration?