 Hard drive data and stats by back blaze back blaze is awesome in terms of that fact they've been doing this So I don't use them as a service, but I've heard they're really good I have considered using some of their backup services But what's fascinating is the peak they give us into how they run their data centers Since 2013 back blaze been publishing statistics and insights based on the hard drives in their data centers And what they actually have is not just the hard drives all the details of what broke inside their data center And they run these quarterly reports and I'll even I'll throw in the links below They tell you their methodologies for testing and they even give you the sequel data for this It's also taught me a lot about things like smart and what smart really tells us about a hard drive is one of their blog posts and The overview of the fact that the smart data is not accurate and not consistent between vendors and we've noticed this too We thought it was always this anomaly some drives said they went on for 10 years But the drive is less than 10 years old telling us that that can't be correct So as we've learned smart is not the most reliable, but here are the statistics now What I like about these stats is they tell you the exact model number for each drive So this is the Q4 report and HST HGST drives Here's the exact model the size the drive count the average age in months that it was running Then they have their drive days drive failures now What's nice is if you see a drive like this one had no failures, but you can also say okay How many we're on there so they've got 46 of this model to shiba with no failures That's a very small sample pool, but when you look at something like this this HM 55. I'm sorry HMS 5 drive They have 7,000 in place And they only had nine so they give you good statistic and you can see we're 446 on this WDC on these new six terabyte ones with five dry failures of 4.4 Failure rate where really it's scary is it looks like Seagate clearly has a drive problem We've seen this before with some of the other reports and this ST series four terabyte one 234 drives out of 34,000 failed which puts their failure rate at 2.67% Pretty high. I've actually used this before I purchased hard drives. You know like I said well We also have a Seagate up here So I'm gonna say not to buy Seagate for terabytes because look at the one up here. This has a 14% failure rate It's only seven drives to smaller sample pool of only 184 But that's still a high number of drives to fail out of that pool But these are great. They have all the details I highly I look forward to this report every quarter and I really think them because you know These hard drives are going in data centers everywhere Backblaze is a little bit different than some of the other companies because they're all these model drives You may notice are not enterprise drives. They actually use a lot of consumer one They have a lot of details they share with how they built out their data center using consumer drives And if you want to run your own statistical values on these right here Literally is the raw data that you can download for the document and data zips Which is a series of sequel files and PDFs so you can Reprocess the data and you know gleam your own insights for it So hats off to backblaze for doing this and if you just kind of wanted some ideas for which hard drives to buy And you don't need the absolute latest hard drives these Ria ability ports, you know They can give you a good estimate and good feel and one of the nice things I like about this is if you see in working in a tech space I see a lot of hard-race fail the problem is I don't have this level of statistics I can to see a lot of people bring a computer in with a certain model drive fail But it was it because that drive was a good price so there's more of them in the market So their failure rates not anymore, but they have higher market saturation versus Or is it really a problem with that particular drive? This is where this helps all that data and you can understand because when you're talking about a drive pool of 34,000 drives Or 1,800 drives or 9,000 drives and they're telling you how many out of that 9,000 drives fail much more accurate much You know more forthcoming with the data I'm really hoping more companies follow them for until more companies start publishing this data We can just rely on em a recorder to release their Details of what drives failed and help us in our drive purchasing information It may be helpful for these companies so they can learn what's wrong with their drives because you know It's a consistent environment so they can try and track it too I'm sure see gates really interested on the fact that their number is associated with 14% failures here and 2.6% failures here out of 34,000 drives Thanks for listening if you like the content here like and subscribe