THIS POST IS A LITTLE OUT OF DATE NOW. WE WILL UPDATE IT AS SOON AS WE CAN BUT MANY OF THE DETAILS REMAIN THE SAME.
Last year, we replaced our entire storage infrastructure. Not really through choice but because a very brief power blip forced our hand. It’s not all bad though as the new storage is nothing but goodness in a little black box.
In this blog post we are going to talk about our own storage and backup infrastructure and why we use what we use.
First, a little background about me. Before I became a filmmaker I worked for a very large IT company as their storage and backups manager. I was responsible for a team of people who looked after petabytes of storage for our various clients. We handled some of the largest NAS and SAN setups you are ever likely to see. While our own setup is considerably smaller we still employ many of the technologies I used to use in my old day job.
For any small business you have a need to store data. For photographers and filmmakers our needs tend to be considerably higher than your average Joe. For us each job can result in over 100GB of data, data we need to store, archive and backup this data. We are going to look at the various options but we will mainly focus on what went for.
NAS VS DAS VS LOCAL
For many the main choices for data storage are to keep it all on your computers local hard disk, use DAS (Direct Attached Storage) like a USB drive or a DROBO or the next step is a NAS (Network Attached Storage).
Local storage is by far the cheapest. Bung a few terabyte sized drives in your computer and you have an abundance of storage for very little cost. Perfect if you work alone and have no need to share or move your data. But what if you want to access that data from multiple computers? You could use a network and leave your PC on but that is a great way to burn a few hundred watts of power for doing very little.
Many creatives opt for the Direct Attached Storage option and work off external hard disks. This gives them the flexibility to work on their main desktop or on their laptop. A step up from this is something like a DROBO which can be shared by a few at the same time but also offers protection from RAID (more on RAID in a moment). The major downside of a DAS is all your computers access it via a USB, FireWire or Thunderbolt connection so they typically have to be in the same room and pretty close. You also have to manually manage any backups and if you want to share files with a wider array of computers you need to leave your PC on. Again, this means burning power you really didn’t need to.
For us, we opted for the Network Attached Storage (NAS) option. A NAS is an external storage unit which connects to your network. It’s basically an all in one server. Just what this server can do depends on which NAS you go for. The HUGE benefit of a NAS is anyone who is on your network, be it a laptop, desktop, iPad or iPhone will have access to EVERYTHING on your NAS from anywhere on your network. With the right NAS you will also have access to all this data when away from the office. A NAS also usually employs RAID for added data protection. Our NAS is the centre of our network. We edit from it, store data on it, archive to it, have our music library on it and also our entire CD and DVD collection.
As a NAS is also a server it can run automated tasks. Ours runs a whole array of tasks such as virus scans and health checks as well as fully automating our backups.
So, if you have the need to access your data from more than one computer or want to automate a lot of your data maintenance then look at the NAS option.
OUR NAS OF CHOICE. THE SYNOLOGY DS413
We used to run a Buffalo TeraStation and I wouldn’t recommend even their latest models to anyone. They are slow, seriously lacking in features and one of the most expensive units you can get. For me, the two main makers to look at are QNAP and Synology. Both major players in the business NAS market and an excellent price to boot. For me, the decision came down to performance. The Synology units offered superior performance per £££ than the QNAP and both packed a similar feature set.
We researched the performance of various NAS units over at smallnetbuilder.net. They have a number of great reviews as well as performance charts where you can compare various suppliers and models. One thing to bear in mind is you really need a Gigabit network to get the most out of your NAS and the theoretical max speed you can get from a Gigabit network connection is 125MB/sec (that’s MegaBytes a second). Some NAS units will offer greater performance but to do this you need multiple network connections and your network switch configured with link aggregation. To do this you will need a rather expensive managed network switch. Your typical small network switch like a Netgear GS608 (what we use) will not allow link aggregation. Still, the DS413 can reach some pretty good speeds. With read speeds around 110MB/sec thats not far off what you can achieve from an internal hard disk. A WD Blue for example will give you around 113MB/sec in your PC. But wait? Doesnt your PC have a SATAIII connection which claims transfer speeds of 6Gbit/sec (600MB/sec)? Yes, that is the interface speed but your hard disk is a mechanical device with a spinning disk and moving read head. The limit of these is usually around 150MB/sec. It’s the marketing guys which will have you think you can hit higher speeds. Write speeds to a NAS are typically lower, especially if you have RAID configured as the units processor needs to do some extra work to calculate the RAID. More on that later.
So the Synology DS413 was our NAS of choice. At the time of writing the best price for the chassis only was £399.99 from Dabs.com. UPDATE: The DS414 is the new model and only £359.99 from eBuyer. You typically buy a NAS without any disks and you fill it with whatever you have lying around. The Synology units feature SHR (Synology Hybrid Raid) which lets you mix and match hard drives from different makers of different sizes and easily configure your RAID volumes. This is a great option if you have a load of old SATA drives you want to make use of. To see what space you can get from your NAS try out the Synology RAID Calculator.
The DS413 is packed full of features. You can configure automatic backups to external hard disks, install one of the add-ons and backup to Amazon Glacier or one of their other providers. Have it perform a virus scan each day, cleanup the trash cans. There are also add-on packages to turn it into your own Dropbox style cloud storage. Stream music to your home entertainment system or AppleTV. You can also use WebDAV to access files remotely via the web. I wont give you a review of the DS413 because there are plenty out there already. Suffice to say, it will server your files fast enough to edit as well as everything else a small office needs.
Other things your Synology NAS can do is run a mail server, offer webmail, disaster recovery data replication, host a WordPress site (our NAS is also our backup website host), handle your DHCP and DNS or share your iTunes library. A NAS can also automate a lot of the admin and maintenance tasks. Ours runs a daily virus scan, checks the disks and cleans house all by itself.
One of the other reasons we selected a Synology is long term support. Usually when you buy a bit of tech you can expect to stop receiving updates as soon as the new version comes out. But like Apple, the Synology Operating System (called DSM) should support your unit for years to come. Looking back the latest version of DSM still supports units from 2009. Here is what the Synology Desktop browser based desktop looks like on our NAS setup.
The QNAP units did come close, especially as they have a package to allow the NAS to directly backup to iDrive (our cloud storage provider) but to get the same speed we would have had to spend quite a bit more. So far I haven’t hit any negatives from using our DS413. It really is a dream NAS unit so far.
In short, the DS413 is fast enough for multiple editors to work from with all the source footage on the unit. When clients pop in we stream our films via an iPad to an AppleTV from the DS413. If we’re out and about we can perform admin by accessing the files back in the office. If you need faster write performance then consider the DS412+.
I’ll get this out of the way now… RAID IS NOT A BACKUP, DON’T BE AN IDIOT AND ASSUME (or hope) IT IS!!!!
RAID (Redundant Array of Inexpensive Disks) is where you make 2 or more hard disks appear like one big disk. For us we use 4x 3TB drives which could give us a total storage of around 12TB. The storage unit does some fancy magic to spread the data over multiple disks and can give you lots of storage or a lovely and robust array which will tolerate disk failures and a disk failure really is your most likely cause of data loss.
This is where you configure your unit to treat all the hard disks as one large volume. In our case it would give us 12TB of storage. The problem with RAID0 is that if one single hard disk fails (which WILL happen) then you lose it all, EVERYTHING! All the data on all the disks in the RAID0 array. Hard disks are mechanical things, they will fail in the 3 years or so you have them. RAID0 can work with 2 or more disks and gives you the maximum amount of storage but without any redundancy and increased risk of data loss. Personally, I never recommend using RAID0.
Using 2 or more disks RAID1 can offer you basic protection but at a huge cost to space. RAID1 is a mirror, what is held on one disk is immediately copied to the other. Ideal for 2 disk units like the DS213 in that if one of your disks fails you lose nothing. The downside is exactly half of your space is given up for the redundancy. Still, it’s better than nothing.
RAID5 (what we use)
This requires 3 disks or more and ideally all of the same size. I wont explain how it works but RAID5 stripes the data over the multiple disks and has something called a parity bit. This means if one of your disks fails then it doesn’t matter, things keep on working. RAID5 also has very economical use of your space. In our setup RAID5 gives us 9TB of usable space with the other 3TB going to the redundancy. That’s just 1 wasted disk vs the 2 we would waste with RAID1. In this setup we can lose 1 hard disk (which we would replace super quick) before we risk data loss. RAID5 also has excellent performance as your data is read from all the disks vs just one.
There are other RAID options like RAID6 and 10 but these are aimed at high performance database servers and cost you more disk space than you really need to sacrifice. For most I would strongly recommend a RAID5 setup for the best performance, space and cost balance.
So I mentioned earlier that you could buy an empty chassis and chuck in your spare disks. This is cheap and a great way to get started but is far from ideal.
Chances are your hard drives are standard desktop class hard drives. Servers and commercial NAS units typically use enterprise class hard drives like the WD RED. Downside is they are around 2-3 times more expensive than a desktop disk and often a bit of overkill for what we need. That’s not to say they aren’t a good idea if you can afford them. What makes enterprise class drives different is they are designed to be on 24/7 so have a much higher build quality and tolerances to things like vibration and heat. They also play nice in a RAID configuration. If your normal desktop drive finds a bad sector (which will happen) then it tries and tries to fix it, by which time your drive could have dropped off the RAID array and make things a little hard to access. An enterprise class drive tries and then gives up.
Western Digital however have a new drive on the market, the WD Red and they are pure awesomeness. They are a techies wet dream. Only a few quid more than a desktop drive but offer the benefits of RAID class drives like the RE4 and are designed for NAS units of 1-5 drives (dont try and us them in a unit with more than 5 drives, they are coded to not work). The biggie is they come with a 3 year warranty vs 2 (you can upgrade this to 4 years plus 1 day RMA for around £10). They also feature increased vibration tolerance and in a unit with multiple drives rumbling around this can really help performance and life expectancy. They also run cooler as they feature a spindle speed of 5,400rpm. Again, multiple drives in one chassis will generate a lot of heat. Our WD Red drives run at around 29ºC vs the Seagate Barracuda in our PC’s which run at 45ºC. WD Red drives also feature RAID friendly error management but the biggest is the performance. Reviews have shown them to outperform even the WD RE4 drives when it comes to small NAS use. Enterprise class drives and the WD Red also feature less aggressive head parking. In a typical desktop drive they are designed to park the read head after 10 seconds of inactivity. In a NAS this can affect performance and drive life and as they are accessing data more frequently is not needed so they typically only park the head when told to power down.
So if your getting started in the NAS world then by all means use desktop drives like the WD Green. They are cheaper but for us reliability was the key and the WD Red was the perfect fit. Their low power draw also means the DS413, loaded with 4x 3TB WD Red drives only draws 23watts when working flat out.
Best price at time of writing was Dabs.com at £118.99 for the 3TB drives.
UPDATE: A 3TB drive is now only £99.99 from eBuyer and they also now do a 4TB model for £134.55. The extended warranty and 1 day RMA is only £6.99.
Here is the biggie, backing up your data. If you are a wedding photographer or filmmaker and don’t backup then you should be ashamed of yourself and apologise to your couples immediately for putting their memories at risk. Go on, I’ll give you a few minutes……
Backups should be mandatory for any business really. Can you really risk loosing anything? Especially a wedding which you have zero chance to film again. If you ever have to say to one of your couples “I’m sorry, we had a break in and they stole all our gear including your wedding” do you really think they will accept that? Your business you worked so hard to build up will be destroyed and worth nothing.
Earlier I said RAID wasnt a substitute for a backup so I will now explain why. RAID protects you against 1 or 2 hard disks failing. Failing as in the drive just stops working. But what if you accidentally save over a file and need to get the previous version back? RAID1 has already copied the new file to the other disk. What happens if you delete something? Have a fire? Have a break in? Do something silly and just need to revert to a backup? RAID is purely protection against a disk failing but there are a million other things that can go wrong with your data, like what happened to us. For us a power blip caused the network configuration file on our NAS to become corrupt. The NAS rebooted and loaded this new configuration into memory, the result was a totally dead unit that couldn’t be retrieved. It wasnt an eventuality we ever saw but it wasnt an issue. We bought a new unit and restored from our backups. No biggie.
Our backup strategy is as follows.
On the DS413 are a few USB ports. Into these we plug some 3TB USB hard drives. We configure the DS413 to perform a backup each night to these attached drives. On one drive it places all of our yet to be edited and delivered projects. On the other are the ones we delivered but are still within the 3 months we guarantee to keep the footage for as per our contract. All other data is handled by our cloud storage iDrive.
These are refered to as our onsite backups and allow us to perform immediate restores should we delete something or make another human error. As they are performed each night it limits our loss to a single days work.
We also have another 3TB drive which we use for offsite backups. After each job we capture the data to the NAS, we then back all the yet to be delivered projects up to this drive which is stored offsite. This protects us against fire or theft but is only updated after every job so about once a week. This is why we have the daily onsite backups.
In summary, here is the protection we have.
RAID protects us against a drive failure, meaning we lose nothing and can keep working.
Onsite backups protect us against a major equipment failure and mean we only lose one days work.
Offsite backups protect against a major MAJOR disaster like fire or theft but we could lose up to a weeks worth of editing… potentially.
CLOUD OFFSITE BACKUPS
Another backup we have is cloud storage. Maintaining external drives is a chore and I don’t know why, but ours usually fail the day before we go on holiday, so we outsource what we can. I mentioned that the yet to be delivered work and archive are stored on external drives. The reason for this is the volume of data is huge and you would need a huge big internet connection to back this all up to an internet based cloud. But we have a lot of data which doesn’t change like our music library, sound effects, smaller documents etc. All of this is backed up each day to our internet based provider, iDrive.
iDrive works by installing a small program on each of the computers you want to backup. So it can be on all of our PC’s and Mac’s as well as your iPhone or iPad. They have plans ranging from 150GB to 1TB (we use 500GB for $149.50 per year). They then backup on a schedule, for us is backs up everything other than the ‘too big’ editing projects. It does however backup the smaller editing project files which are only a few MB. So the onsite and offsite backups handle the large stuff and the cloud based storage handles the smaller stuff.
With our setup we could have the entire office blow up and we would only loose a few hours of work thanks to our combination of offsite hard drive and iDrive cloud storage.
If you backup weekly you have to ask, can you really afford to lose a weeks worth of work? How much would that cost you in time? With a NAS unit it’s not like backups are any effort, they take care of themselves and alert me to any failures.
You can give iDrive a try with a 5GB free trial or use this link for iDrive discount code and get 50% off (we get nothing for you using this, it’s just a great discount).
Another cheap option for archive and backup is Amazon Glacier. They charge just $0.011 per GB/per month but they do charge for each upload, download, access and bandwidth. The benefit of Glacier is the Synology and QNAP units have a client so the NAS can backup to these directly.
There are a whole host of cloud storage suppliers like Carbonite who offer unlimited space but we found that only iDrive would allow us to backup mapped network drives like our NAS. All the other were strictly limited to the local PC.
ADD A UPS FOR THE ULTIMATE SETUP
I mentioned that the reason we had to spend £1000 on a new unit and disks was because of a power blip. Literally one of those that turns the lights down but doesn’t even reset your microwave clock. Hard drives are mechanical with a little head (like a record needle) which overs over a shiny magnetic disk. If the power goes suddenly this head crashes onto the disk surface and causes physical damage. This is why you are always told to not turn off the power while accessing your hard disk. A UPS (Uninterruptible Power Supply) is not necessary but they can save you spending £120 on a new hard disk or worse yet £500 on a new NAS. Better yet they can avoid corruption of your valuable data and they cost from as little as £80.
Another reason to consider a UPS is to avoid data loss when writing. When you are writing to a hard disk you don’t actually write directly to the disk, that would be waaayyyyyy tooooo sloooooowwwwwww. What actually happens is it writes to the disks cache and then the cache is written to the risk. In the event of a power failure the disk is supposed to use its remaining momentum to complete the write before the memory is lost. This doesn’t always happen and is more of a sales pitch than an actual thing. For this reason a UPS can help as it allows all writing operations to finish before shutting down the disk safely.
A UPS is basically a car battery which can power your computer and other tech gear for 5 minutes or more. The more you spend on a unit the more standby time you get.
We went for an APC Back-UPS Pro 550 from eBuyer for £104 model number BR550GI. This unit gives you 3300watts of power and can power our NAS and other networking gear in the event of a power failure for over 3o minutes. More than enough time for the NAS to power itself down safely and wait for that glorious electrical power to come back. Where it then turns itself on again. This unit could power your Mac/PC and storage for 9 minutes or more. It has a USB interface so can talk to your NAS or computer and issue commands to safely shutdown. The battery will last about 3-5 years before it needs replacing but can all be done by the user for around £30.
So we have our NAS as our central storage unit with all our data and everything backed up. What this means is on our computers we have nothing but the operating system and our apps. I can say hand on heart that if our computers were to blow up or have a disk failure then we would lose nothing. When I hit save on an editing project its saved to our RAID5 array so we are protected against disk failure. Around lunchtime that will backup to the offsite cloud storage and at night it will backup to the external drives.
We upgraded our computers earlier this year and I didn’t have to spend a day backing things up or copying files over. I just built the new machines and off we went.
I’m a little rusty but here is a rough diagram of our network setup.
A NAS is the ideal solution to allow the best use of your storage and share it among all of your users. Why give everyone a 3TB external drive and have them only use a small portion? Share the love and the storage. They are also so simple to setup. The QNAP and Synology units are designed to be used by the home user novice. Throw in some WD Red drives for the most reliable and speedy setup you can get.
A NAS will also handle all your backups and other data management tasks. They will tell you when a drive is going to fail and if one does it doesn’t matter. If you use RAID5 you can keep running and keep editing.
Backups are essential, especially in our field. We don’t get any re-shoots or second chances so your backup strategy should be rock solid. Cover yourself for EVERY possible eventuality with regular onsite backups and at least weekly offsite. No one needs backups, there like insurance. But when you do, your glad you have them. What can go wrong and should you cover for? Drive failure, storage unit failure, deletion, corruption, user error, fire, theft, squirrels.
A UPS is by no means essential but it can save you time having to run data integrity checks after each power blip, and for just £80 why not?
And remember, you’re a business. Whatever solution you go for should be within your budget and reflected in your pricing. Our bullet proof strategy is all part of our premium service. If your thinking “I can’t afford backups!” then it’s clear you’re not charging enough. The cost of data storage and backups for each of our clients is approximately £30 if you take into account the cost of the unit, depreciation over 3 year, ongoing maintenance and running costs. That’s assuming we keep the data for 3 months after delivery.
Have any questions? Then feel free to ask them in the comments below and we will do our best to answer them.