I suppose it’s pretty clear that optical discs are no longer the favoured medium of choice for many applications now, however, there still remains a sizable community that use Blu-Ray recordable discs as a form of storage. This may be because of the need for mechanism-independent, read-only storage or the need to have it as a vehicle to secure, transport and reproduce high-definition video features.
Those who have been following may have already realized that the use of optical discs represents a big uncertainty, as the materials used to create the disc can degrade over time, putting the data stored on them at risk. We have already seen this happen with CD, DVD and also in my case, Ritek Blu-Ray discs.
In order to try and get some data about the longevity of Blu-Ray discs, I have run informal experiments on a select sample of my own archive collection. These tests are, and cannot be, perfect and a long-form discussion follows the scanning results. However, as these samples have been established prior, it forms a good baseline for us to do some comparison – on the absolute scan values, and on the relative scan values to other media-IDs and to their own results in prior scans.
The raw data can be pretty difficult to appreciate on its own, so I will do my best to try and make the important points up-front.
The methodology of disc scanning has subtly changed over the course of my “scanning” career. The methodology from this round of testing is as follows:
- Opti Drive Control version 1.70 running on Windows 7 Ultimate x64 (upgraded from Opti Drive Control version 1.51 running on Windows 7 Ultimate x86 due to hardware failure). It provides some bug fixes however should not affect the scanning accuracy. The test machine is a HP Microserver N36L in both cases.
- Lite-on iHBS212 Blu-Ray rewriter drive, same as used before. The firmware has been kept the same at HL05. I had previously established that the iHBS212 is harsher than the iHBS312, although following the same trend, so I have kept the same drive for the reasons of consistency.
- The drive itself has not been used for reading or writing since the last test run, and should not have received any significant wear and tear that could affect test results.
- Scanning speed has been standardized at 4x, upon the request of the community and upon the results of my prior study into scan speed.
- Concerns about drive heating have been addressed with a 30 minute or longer break between consecutive scans.
- Room is maintained at 22 degrees C by an inverter air conditioner. Discs and computer have acclimatized in there for several days.
- Concerns about disc cleanliness have been addressed through a cleaning regime – using tissue, high purity isopropyl alcohol is rubbed on the data side until it is covered with drops/pools of it. The disc is then inverted, data side down, onto a fresh soft low-lint microfibre cloth which has not been touched by fingers. The disc is then rotated and rubbed around the cloth until the data surface is dry. The disc is quickly turned over, inspected with a high intensity LED torch to verify no fingerprints or oil residue, or smudge/water marks remain before quickly transplanting it into the drive. As a result, present scan values may return lower error rates than before.
- To be consistent, clean discs which did not need cleaning have been cleaned as well.
I have done the above to ensure the test results are as valid and as useful as possible, however, I anticipate that somebody somewhere will have a problem with it somehow. Unfortunately for them, this is the method I have settled on.
(Click/open in new tab for larger, readable! Colour coding represents the value compared to the other values in the set. Some values are consistently good throughout will be seen to be green mostly, although factor in the age of the disc!)
- In testing, all RITEKBR1’s have failed. Previously, two in twenty were readable, now all have failed.
- None of the other discs were unreadable in drives, despite elevated error figures. I suspect that our testing limits of LDC Average <=13 and BIS Max <9 are too stringent based upon the unreadable scans of the RITEKBR1’s. Suggest that BIS > 50 is a cause for concern, but not directly implying unreadable (TDKBLDRFB achieved BIS peak 87 and was still readable!).
- The number of discs that strictly met the above tested criteria are very few – just one VERBATIMc as far as my eye can see.
- Degradation or inconsistency seems to plague TDK Media for some reason.
- Verbatim discs provide good results all round, age and absolute values being considered. I would highly recommend them of the media codes I have tested. MEI discs are also highly commendable based on their consistency as a dual layer disc.
- Physical degradation of the LG media was noted in an earlier posting, it is now clearly visible and will soon threaten data similarly to the other TDKBLDRBA sample 2 disc.
- Transfer Rate Test (TRT) and CDSpeed Scandisc tests are highly insensitive at picking up low-level errors which appear in these scans.
- Some discs have seen significant scan level result benefits from the new cleaning regime, although the benefits are occasionally non-uniform due to the previous conditions of the discs being non-uniform.
- New media have been added to the test pool, although I can’t find my VERBATIMv samples, so they’ll have to wait (error in the chart, should have said VERBATIMv not VERBATIMe).
- The lifetime of the early-generation 2x and 4x Blu-Ray discs seem to be limited, and we may be seeing failures within the first 5-10 years of life, if not already (e.g. PHILIPR04, RITEKBR2). The longevity remains a cause for concern, if not using the premium media. Unfortunately, premium media does not compete well with the cost of magnetic hard-disk based storage at the present moment.
- The discs all contain data and do not have a consistent size. This will put the larger discs at a disadvantage, as the outer edge experiences elevated error rates first in most cases (but not RITEK failures in my experience!)
So, as usual, it’s not perfect. But none of my discs bar the RITEKs have failed outright. In some sense, this non-eventful report is exactly what you might be looking for.
Individual Results and Discussion
TDKBLDRBB Sample 1
This is one of my earlier burnt discs, and the result is rather astonishingly poor at the outer edge. I thought it was my bad cleaning job, so I gave it another rub and tried it again.
Only slightly, but almost imperceptibly better. It proves that my cleaning isn’t hurting anything though, which is a good sign. It’s hard to avoid scratches and spreading oil when doing rigorous cleaning. The repeatability and signature of the disc is present in both scans from the shape of it.
TDKBLDRBB Sample 2
This one fared a little better, but still not as well as I would have liked. The error rates had not elevated as much as the last one, but it’s still elevated. The differences may be because these discs came from three separate TDK branded batches.
TDKBLDRBB Sample 3
This one fared the worst of them all, but a TRT proved that the disc is still completely readable without slow-downs. It suggests we may be too conservative with our error-rate margins.
TDKBLDRBA Sample 1
So, while the 4x editions of the disc above didn’t fare so well, this 2x disc turns up a surprisingly good result. The recording layer is different, being more a brown-blue rather than brown-brown type. The sad thing is that the partner disc in the pair had failed due to oxidation of the reflective layer, and so it implies inconsistent manufacturing which directly affects the disc’s longevity especially when handled or exposed to high humidity.
VERBATIMc Sample 1
For discs of a similar age, the Verbatim puts out a pretty good scan. These were overprint Verbatims, which are likely to be of a lower grade than the “real thing”, yet still they perform admirably, this scan being better than the scan recorded prior (likely due to dirt being removed by the cleaning).
VERBATIMc Sample 2
Alas, this one is not quite as good, but still is a great performance. I suppose to keep in mind that these discs were burnt at 6x (their rated speed, and the maximum speed of the LG GGW-H20L), one cannot expect as good of a burn quality than at 4x CLV or 2x CLV.
CMCMAGBA3 Sample 1
CMCMAGBA3’s have been the “go to” media for price and reliability since the RITEK fiasco. Unfortunately, it seems that the CMCMAGBA3’s have seen elevated errors on the outer edge as well, similar to the TDKBLDRBBs. I’m not too happy with that, but it’s still readable.
CMCMAGBA3 Sample 2
The second sample shows a much less prominent rise, the slight rise being a “regular” feature of most discs as the mechanical rigidity of the disc and the tracking demands on the drive are more difficult towards the outer edge, and it is more susceptible for contamination.
CMCMAGBA3 Sample 3
This one, again, shows a fairly marked rise in the error rates towards the outer edge. I think it’s a sign of “you get what you pay for”.
PHILIPR04 Sample 1
The PHILIPR04’s I have are rough edged, and seemed to do fairly well initially. Needless to say, the error rates all around seem to have popped up somewhat, rather evenly, and it’s not that good of a sign. It’s still not dangerous, but as some have commented on their failure, I will keep an eye on them.
PHILIPR04 Sample 2
Again, a very similar pattern is emerging. Considering these discs are half as old as the TDK’s, it’s a bit concerning.
LGBRAS04 Sample 1
LG discs have always scanned fairly decently, and this is no exception. It has held up fairly well, in fact, both LG discs have when it comes to error rates. The increases are fairly small, but the disc gives a visual sign of its distress, similar to the TDKBLDRBA’s.
In this case, it’s fairly slow and hasn’t reached the data area just yet. But it’s only a matter of time before the data at the edge is at risk.
LGBRAS04 Sample 2
Unfortunately, as a result, even though the error rate looks fairly consistent (although elevated), I cannot recommend these discs for long term storage either.
INFOMER30 Sample 1
If I must say anything, I bought these discs as a joke but they’ve been surprising me from day one! They seem to be holding on somewhat well, despite their nasty reputation in general. They are, however, just one year old. To see the increases number in the 20-25% of total counts may imply the disc has a lot of degrading to come really soon.
INFOMER30 Sample 2
Again, a superlative result from such a nasty disc. We will continue to monitor and see if this can be maintained.
MEIT01 (DL) Sample 1
These dual layer discs have always been a bit of a concern for me, as error rates near the layer break often escalate quite strongly, but this particular disc seems to shine. MEI stuff is good stuff! Previous results for this particular disc seem to have been affected by dirt.
MEIT01 (DL) Sample 2
This is more what we’re used to seeing, although it’s not the worst I can imagine.
MEIT01 (DL) Sample 3
This one’s a bit messy, there may have been a speck of dirt still remaining resulting in a spike in both layers, but the layer change is surprisingly clean.
TDKBLDRFB (DL) Sample 1
This is one of my fears when it comes to dual layer discs. This shows the same kind of performance that the other TDK discs do. Maybe it’s just a fact that the LiteOn drives have some interaction with the TDK discs that cause the error rates to be amplified, as I did a Scandisc on the LG and save for some yellow blocks (which are not read errors), it is perfectly fine.
TDKBLDRFB (DL) Sample 2
The same thing holds true for the second sample, and to a more extreme case here – a peak of 80+ in BIS is not common to see, but it’s still perfectly readable in the LG …
So, one has to ask, what is really going on – and what is “readable” and not?
RITEKBR1 Sample 1
I had two Riteks which I had kept of an initial twenty, those were the last two that were readable. I decided to keep testing them to establish the limits of readability. See that above? That’s what happens when a LiteOn “officially gives up”. Sometimes, you can get an answer, if you try several times …
The two error spikes overshadow the rest of the scan, and the jitter values go funky, but it seems that Ritek degradation starts near the centre of the disc. That’s in distinct contrast from what we know and expect from everything else that’s around.
Even with such poor results, guess what the LG can make of it?
Only two stretches of bad/damaged blocks. It seems Blu-Ray can be more resilient to read errors than the Lite-On scans would imply. So maybe we should sleep more soundly at night even when the scans turn a little orange.
RITEKBR1 Sample 2
Sample 2 was a little more stubborn. I could never get it to scan completely at all.
The disc itself was in quite bad shape with long red-block stretches. This is officially the end of the RITEK scans since 0 of 20 discs have survived a readback test. The 100% death mark is reached in just 5 years, with a fraction failing in just 1.5 years.
All of these media were burnt this year, and is expected to have suffered minimal to no degradation whatsoever. These discs will be used for reference in future years, although the lack of data burnt to some of these discs will make direct comparison and advanced failure warning a little more difficult.
CMCMAGBA5 Sample 1
I criticised this media code for providing poor out-of-the-box burn results, but I think in comparison, that if it doesn’t degrade, this could still be an acceptable result. Unfortunately, as I wasted almost all of my media in tests, I only have one sample reference disc.
PRODISCR0 Sample 1
This has been my “cheap” batch of media which I use from time to time. It looks like it’s okay except for some outer degradation which may be burn strategy related. It’s not as good as some of my other results with the media, but it is sub $1/disc.
PRODISCR0 Sample 2
This one seems to have a few “spike” patches, I’m not sure why either. But I’m not expecting wonders of longevity and reliability from Prodisc (if my previous experiences with their other media are anything to go by).
TYG-BDY05 Sample 1
Taiyo Yuden is a name which we have come to love and cherish. I suppose this result isn’t surprising, although the spike certainly is. This was one of the test discs used in initially evaluating the media, it’s an LTH so I can’t expect the longevity to be great.
TYG-BDY05 Sample 2
Again, a little spike, but worse quality overall because it was burnt in an iHBS312 at 6x (rated speed) and the Lite-On burn quality just doesn’t seem to be up to it.
UMEBDR016 Sample 1
Another media with a mixed reputation that I got just for a bit of experimenting. I initially had good thoughts about it, and its underside also has an ashen grey somewhat in common with the Verbatim. I was hoping it would be better, but it seems to have some elevated errors.
UMEDBR016 Sample 2
This one is not so bad but the spikiness is really what bothers me.
Seeing no additional failures in my testing was a good sign, although elevated error rates continue to be witnessed. The boundaries for readability seem to be much further than our guidelines presume, so we can probably sleep a little more soundly in regards to whether discs are failing or not.
It’s really hard to justify continued investment into Blu-Ray due to the time, cost and uncertainty factor. I really only recommend using it where it is the only option (e.g. burns of BDMV data).