Posts tagged backup

The SSD Swap

A while back, my wife’s PC decided to throw up a couple of ATA errors. ATA errors are usually associated with disk issues so I took them very seriously. A series of disk checks reveal a couple of bad sectors. For me, a bad sector is a indicator to replace the disk ASAP. As this was the boot drive, my first priority is to update the backup image. I use Acronis TruImage to create backup images of my boot drives. Data drives are backed up to my NAS (1st level backup) which automatically replicates to my Storage Server that houses 6x2TB of disks.

After updating the image, and had chkdsk mark up the bad sectors, the problem went a away. Now this is a temporary solution as a bad sector usually indicates something is wrong or is going to go wrong on the disk. It usually never goes away. It’s like a tumor that even though you’ve beaten it into remission, it can always potentially rear its ugly head. Of course, if you cut the tumor out, it’s a different story. So, the analogy was to swap the drive.

I contemplated swapping with Seagate’s hybrid drive but decided to try out Corsair’s F60 SSD instead. It’s a Sandforce powered drive that’s been reviewed to be among the quickest at the time of writing. I went with 60Gb as it was an affordable size. SSDs are faster in larger sizes but any SSD will beat most mechanical disks anyway so…

The result, well, cold boot to usable desktop (i.e. I can launch Firefox or Photoshop) went down from about 119 seconds to about 31. This is Vista 64 by the way. Application launches went from 21 seconds down to 4 seconds for Photoshop CS4, Firefox takes about a second to pop out. Outlook took about 6 seconds (not too bad considering all the offline and email folders are on a mechanical disk).

Overall, definitely very, very snappy which is the main point about SSDs. It makes my other systems boot time feel like forever. Even shutdown times are under the five second mark.

Well, my next SSD upgrade will probably be a Sandforce 2 powered drive. I can’t wait!

Storage & Backup again

Seems like there’s a raft of people I know who recently lost some stuff due to disk crashes or are looking at ways to store that ever growing pile/stack/stash/gigabytes/terabytes/petabytes of data that people nowadays are generating. Even non-photographer friends I have are either busy downloading TV series or accumulating lots of media here and there such as e-books, songs, photos of their children, etc. Generally, media is usually the biggest culprit. I mean how large can a collection of excel spreadsheets or word documents be?

My approach is a two-tier storage system with my QNAP NAS playing primary on-line storage and a custom-built machine serving as the full storage backup. The following is a simple sketch of my network with my QNAP and my storage server. They’re connected to my Cisco switch using Link Aggregation (802.3ad) which offers both redundancy and faster transfer speeds. The little diagram on the right is how my content is stored between the server and the NAS. In a nutshell, my main storage server contains everything but the NAS has a little sub-set of the data.

Basically, my server stores everything in 1TB & 1.5TB disks, mirrored of course. I use rsync/robocopy to mirror as I’ve mentioned much earlier about using cheaper RAID cards. The mirrored pair is exposed on the network with one disk primarily for reading and the other for writing. What basically happens is that I back up to my NAS and my Storage Server (when it’s switched on – which is like at least a few times a week), pulls the backup from the NAS. When working on photos (especially for paid clients), I usually back-up simultaneously to both my QNAP and my Storage server. You can see below where I have my disks…

In case you are wondering (which you will), my disks are all short-stroked. My two swap disks are 150Gb 10,000rpm Raptor disks. Also, if you noticed, my storage server seems rather short of space… well, that’s because all the files are using NTFS junction points since I don’t have a RAID array… there’s actually 3 sets of mirrored disks. Total capacity is 1TBx 2 and 1.5TB x 4 / 2 due to the mirror. Usable capacity is about 3.4TB. Using junction points, 2001 and 2002 directories can reside on Pair A while 2003 and 2004 can be actually on Pair B. A bit complex but cost wise, all you need is a system with more than 4 SATA ports and you’re set. Recovery is also easier. Currently, the Samsung HD154UI 5,400rpm 1.5TB disks are good value but I’ll be moving away from them considering their slow-as-snail spin-up time and generally overall poor performance. I’m testing Hitachi’s affordable (runs a little hot though) 2TB, 5-platter, 7K2000 drive and see if it’ll give me back some performance lost to the Samsungs. Green drives may consume less power but the slow speed may result in high power consumption if we end up leaving more systems on just to complete back-ups and disks synchronization.

So far, I’ve been satisfied with my Seagate 7200.12 1TBs but looking for 1.5TB and 2TB drives are tough if you’re trying to avoid the green marketing crap.

Storage strategies

A few people have asked me to elaborate on my storage strategy so here it is. Some background information is needed first. I started taking photography seriously in 1998. Buying my first SLR back then was a big thing and working through the lenses and other options. In late ’99, was basically scanning the prints on a flatbed scanner. Quality was amazing, or at least what I thought was amazing until I stumbled upon a very expensive (at that time, I wasn’t making much a month) option; the slide scanner. By skipping the printing stage and scanning directly from film, it delivered unmatched results from anything else unless you’re comparing to a drum scan. Since then, I’ve been scanning almost all my slides and negatives but most of what I shot were from my travels. Film costs a fair bit, especially if you shot transparency film such as Fuji Velvia, Kodachrome or Elitechromes, so the everyday life doesn’t often get captured on film.

So with digital scans dating all the way back to 2000 (the 2700 dpi scans tend to produce on average 14MB LZW compressed TIFF files), storage is key, along with a proper backup strategy. I went with CD-Rs and then DVD-Rs but they never were convenient (not fun, not quick, and laborious process) meant that you will almost never will do it timely enough so I decided to stick with hard drives. They get cheaper and are inherently quicker to access if they’re stuck into a spare machine that can be booted up anytime you needed to.

I upgraded scanners so the file sizes got bigger and films like Velvia and EliteChrome 100 pushed the resolution (grain?) further. Then came my first DSLR, a 300D (my shortest-lived DSLR ownership) and my 20D several months later. So came RAW files, 16-bit PSDs, and the story continues until I hit my 5D Mark II with 21-megapixels, Full HD movies. Backup is getting more important but also important is accessibility.

I needed to backup and offload my main PC but also need to access the “older” stuff occasionally. Leaving your main storage server switched on/off daily or 24×7 wasn’t something I want to do on a daily basis so I got a NAS which I quickly outgrew within a year so that too got upgraded to a 4-bay unit.

I have a storage server with 6x1TB disks. It’s all RAID 1. Mirrored. So I get a net space of 3TB. I have a “_photos” folder shared out that further breaks down by year, i.e. “2001”, “2003”, along with some odd directories. I also have a “_videos” folder where I back up my created DVDs (ISO format) and Adobe Premiere Pro project files. The problem is occasionally accessing stuff from say, 2006. My main PC only stores past 16-odd months so I ended up going to my storage server regularly. So what’s running 24×7 is my NAS box with 4x500Gb in RAID5. That gives me about 1.4TB of disk space. That way, I get that past 2-years on it; i.e. 2008 & 2009 along with my 60Gb music collection on the built-in iTunes server. My IP cameras also record to NAS and they keep the past-7 days recording. I also have imaged backups of all the machines in my home so that if the primary disk crashes, recovery is just a click away. Keeping everything in sync is the unix utility rsync (the windows equivalent is DeltaCopy).

With the growing file sizes, I’ll be moving my storage server to 1.5TB disks by Q4 this year and my NAS will grow to 4x1TB*… one can never have enough disk space!

* That’s one thing I like about my NAS. I can live swap the disks and expand it without powering down and losing my data. I can easily move from 500Gbx4 to 1TBx4 and later to 2TBx4.

Disk Failure rate…

After I mentioned on Facebook and Twitter about my recent Western Digital hard drive failure, someone asked me what are my other disk failures. So happens, I do keep records of my disk failures. So here goes, all the way from 2003.

2002 : Nov : Quantum Atlas 10k II QM318400TY-LW, 10,000rpm USCSI3. Some bad sectors, recovered from spare sectors.
2003 : June : Western Digital WD1200JB – 120GB PATA, 7200rpm. Disk crash after bad-sectors detected.
2004 : Oct : Western Digital WD2000BB – 200Gb PATA, 7200rpm, Bad-sectors, discontinued using
2005 : May : Seagate Barracuda 7200.7 (ST3160023AS) 160Gb SATA, Bad sectors, Warrantied, replacement retired.
2008 : May : Western Digital WD2500JD – 250GB SATA, 7200rpm. Bad sectors, retired
2008 : Aug : Western Digital WD5000AAKS – 500GB SATA, 7200rpm, Bad sectors. Replaced under warranty
2010: May : Western Digital WD5000AAKS – same drive above. Retired.

What drives do I have running on my desktops, server & NAS right now:

Hitachi Travelstar 5K100, 5400rpm, 100Gb, SATA (2.5″) – 1 drive
Seagate Barracude 7200.9, 7200rpm, 500Gb – 3 drives
Seagate Barracuda 7200.11, 7200rpm, 1TB – 5 drives
Seagate Barracuda 7200.12, 7200rpm, 1TB – 2 drives
Seagate Momentus 5400.2, 100Gb, SATA (2.5″) – 1 drive
Samsung SpinPoint F3, 7200rpm, 1TB – 1 drive
Western Digital WD10EACS, 5400rpm, 1Tb – 1 drive
Western Digital WD5000AAKS , 7200rpm, 500Gb – 1 drive
Western Digital Raptor WD740, 10000rpm, 74Gb – 1 drive
Western Digital Raptor WD1500, 10000rpm, 150Gb – 1 drive

Guess what brand I’ll stick to for now?

QNAP TS-439 Pro unboxing and mini-review

The big box

If you’ve been following this blog for a while, you’ll have noticed a few posts alluding to the search of some network storage to replace my 2-drive unit. I finally settled on the QNAP TS-439 Pro. Read the previous post about it or go to QNAP’s site to read more. Basically, I needed a compact, four-drive unit. Why four drives? It allows either two mirrored sets of disks (note: not necessarily RAID1) or RAID5 (with or without a spare). Going larger (5, 6 or 8 drive units) adds exponentially to the price but I wanted the use of the few 500Gb disks I have lying around. 500Gb is quite small but if I put them in a RAID5 set, I get 1TB (n-1). Later when I move some of my main storage disks to 1.5TB disks, I get another 500Gb to slot in as a spare drive (3x500Gb + 1 spare).

Other highlights for me were the hot-swap disk trays (lockable), multiple USB ports for printer sharing and dual-Gigabit LAN ports. The other stuff are frills. The dual gigabit Ethernet ports give you a fair bit of flexibility. You can configure two separate links for two separate networks, or some form of load-balancing, etc. Mine is configured for link aggregation group (LAG) mode. If you want to read more on that, which can get technical, Google for 802.3ad. In a nutshell, what LAG does is to create an aggregation group of network links that share the same speed and duplex settings. This requires a switch that supports LAG though and few consumer grade switches support it. The TS-439 Pro does have another option of using Adaptive load balancing. This bonds the network interfaces on the 439 and allows it to aggregate without any switch support. There’s a slight drop in performance but if you don’t have a switch that supports 802.3ad, it is the way to go. Did I mention two links means redundancy?

Pretty basic and unassuming, low power, low heat and quiet.

My primary use is to serve as my iTunes and media server (uPNP/TwonkyMedia), download stuff (BitTorrent), IP camera recorder, and a common file store to use between my PCs and notebooks in the house. My plan is to copy my photos there from camera, then once they’re sort of done (processing, etc), they can be moved to my File Server (which has 6x1Tb) in a RAID1+0 config. I’ve used jperf to do some testing to check the LAN bandwidth/switch speeds and for the TS-439, some regular file transfer timings.

First off, my gigabit LAN between Win Server 2003 R2 (dual-core Athlon, 6x1TB, 3Gb RAM, single link, the onboard NIC is died earlier this year so I’m running an Intel Gigabit CT off the PCIe x1 slot) and my Vista 64 SP2 client. I ran jperf over a few minutes with a five second interval. Min speed reported at 601,274Kbits/sec, max speed reported at 879,761Kbits/sec. Average the 30 readings and I get an average of 795,208Kbits/sec, or 776Mbits/sec or about 97Mbytes/sec. So let’s see how the 439 Pro transfers — remember, I have link aggregation on the TS-439.

1. Transfering 14,521Mbytes of photos. What I have is a mix of small and large files sizes. Large as in on average 26,912kb and small as in about 9kb (the XMP sidecar file). Transfer took about 320 seconds. 45.3MBytes/sec.

2. Transfering only the CR2 files took faster, clocking in about 261 seconds, or 55.5MBytes/sec.

3. Dual client access. Transfer a directory of photos and sidecar XMP files (12,672Mbytes and 11,760Mbytes). Start the transfer on client A then client B. Same target on NAS. Client A finished at 262 seconds. Client B finished at 302 seconds. Note that client B started transferring at roughly the 30 second mark of Client A. Average transfer speed is about 48.3Mbytes/sec for client A and 38.9Mbytes/sec for client B. Not too shabby considering both were hitting the NAS at the same time for about 87Mbytes/sec. Good enough for me.

Testing single disk setup

I’ve used it as a mapped drive for working on my photos and from Lightroom 3 beta 2 and Bridge CS4, the images feel very snappy. Not as quick of course compared to my local SATA drive but it’s not that obvious you’re working through a network. The lag on bridge in terms of refresh for example is noticeable only when you have three clients hitting the NAS but the fact is that its not terribly slow. My guess is that if Bridge refreshes my 800 photo directory on my local drive in 5 seconds, it takes about 8 or 9 on the NAS and perhaps 15+ seconds if you have multiple clients hitting F5 at the same time. Overall, I’m pretty satisfied. Maybe it’s time to look for dual-Gigabit port NICs for my client machines as well but I think the NAS will get faster when I up the RAM to 2Gb.

Fancy cover flow style admin interface

Overall, I’m pretty happy. It serves up my music to my iTunes, I can also listen to music/look at photos/watch movies on my Popcorn Hour C-200 via Twonkymedia server, record from all four of my IP cameras and work on a shared folder of photos. Not cheap but probably more cost effective to run than a standalone PC or server (if you factor in licensing, power, heat and noise). You can do other things with it as well though even thought I don’t such as running a Web or FTP server, etc.

If you want a more in-depth review, you can read it here.

Most administrative interfaces are web based now…

Network Attached Storage

From Wikipedia : A NAS unit is a computer connected to a network that only provides file-based data storage services to other devices on the network. Although it may technically be possible to run other software on a NAS unit, it is not designed to be a general purpose server.

NAS provides both storage and a file system. This is often contrasted with SAN (Storage Area Network), which provides only block-based storage and leaves file system concerns on the “client” side. SAN protocols are SCSI, Fibre Channel, iSCSI, ATA over Ethernet (AoE), or HyperSCSI. Despite their differences, SAN and NAS are not mutually exclusive, and may be combined as a SAN-NAS hybrid, offering both file-level protocols (NAS) and block-level protocols (SAN) from the same system. Considering I’m writing for home-users, a NAS makes more sense.

With the amount of data many photographers (and users) are generating, it’s become obvious that storage is a problem. For casual users, it probably may not be a problem in the beginning but it soon will be. I know of friends who after a few years, a kid or two and the amount of photos, videos, music, documents, scanned PDFs, etc start to add up to gigabytes. Burning to DVD (gosh, who burns CDs today?!) can alleviate some of the problem but with 4.5Gb (dual layer disks aren’t common as they’re not so cost effective), it’ll take a few DVDs for the average persons’ backup. Worse still, those DVDs may not be readable after a few years. A friend of mine burns two discs for each copy — and 3 years later found out that some are not readable. Good thing he had two copies as he’s managed to combine them to even out things so no loss occurred. So, he consolidated the storage. Where? Good old spinning platters. With the falling prices of hard drives, it’s a no brainer that disks are still the most reliable, cheap and effective storage for backup for most users today (most users here I mean home and small business perhaps). Reliable? Yes, if you have two disks and you replace them religiously every 36-months. That’s what I do. The benefit is that you’ll end up having larger disks too. My backup server has 1TB disks now. It started out with 160Gb and 250Gb disks (two pairs) for a capacity of 450Gb. My data very important to me so I change disks every two years. I then went to 320Gb and 500Gb (still two pairs) and now, I’ve got four 1Tb disks plus a pair 500Gb disks (these store my music, video and misc files). They’re going to 1.5TB disks soon (best price per Mb).

The problem is usually, you won’t want your backup server running non-stop. Mine’s on a UPS along with my network, and more but most casual users also don’t need access to their backups every single instant. They also normally don’t have another spare machine lying around. In many organizations today, they use the term near-line storage for things that can be retrieved faster that tape but slower than current online storage. Of course big organizations use fast disks arrays for online and slower and usually SATA disks for near-line storage. In my case, I’ll call my storage server near-line because it may not have been started up. Where’s my online storage? Well, it sits on a NAS, a QNAP TS-439 Pro. This is a bit too much for most home-users (they can go with a TS-210 or TS-219) instead. I’ve a RAID 5 set of 500Gb disks in there, giving me just under one Terabyte of available storage. It makes a good always on file-server, repository and immediate backup (Downloader Pro immediately makes a copy of my photos there). When time comes, it gets rysncd to my near-line storage server. The reason I got the 439 Pro was because it was on promotion. It’s been replaced by the Pro II (better CPU) and was the same price as the 419P. To me, the benefits were the faster (compared to 419P) CPU, and larger RAM (1Gb 439 Pro compared to 512Mb on the 419P) which translates to faster transfer speeds. The 439 Pro also has the memory on a SODIMM which means I can bump it up easily to 2Gb. QNAP’s disty over here still has a few units of 439 Pro if anyone is interested.

So what’s for the average home user? Well, if you have more than one computer in the house, would like to share some storage, do some backup, stream iTunes around the house, etc, then any one of these NAS units over here, here and here would serve you well. You’ll also notice that I’ve avoided any form of single drive device. That’s because most home users “think” that once it’s backed up, it’s going to be safe. Two drives are always better than one. If you’re on a tighter budget, get a pair of USB drives. Seagate or WD are good options. Just buy two and make sure your data is identical on each one. And, yes, please store them in separate places — at least a different room. My storage server and NAS are three walls and fifty-feet apart. Note: RAID does not prevent idiotic errors like deleting your only copy of the most important file or photo you ever have — the array isn’t going to care and will promptly erase all for you. In my case, well, I can still go back to either my storage server (100% read-only) or desktop.

My choice of 4-bay, well it’s more for RAID 5 as I’m using older disks for it (i.e. the ones retired from my storage server) so it’s good to have RAID5+1. RAID 5 capacity is (n-1) resulting in the available space of two disks, i.e. in a RAID 5 set of 3x500Gb, I get 2x500Gb usable. The +1? That’s the hot-spare in case something dies which is for the paranoid like myself. If you go with plain RAID5 on my setup, you’ll have 3×500 resulting in a usable 1.5TB.

I’ll post again why I chose the QNAP NAS plus a mini-review… I’ll stop here… more to come while I leave you with a pretty picture of my network appliances 🙂