New BackBlaze Data is Out
-
-
So they are nice disks but expensive and still die.
They must have a massive ROI time period... if they all survive that long. -
I'm still using 1TB Harddrives for my video production. I find that anything higher the performance does down some (due to seek time I guess). And I've been using the same 7200rpm and 10krpm drives since approx 2008 when I bought them. Heavily used with maybe 1 failure.
-
@thecreativeone91 I take it you can't justify making a SSD machine with enough storage to work on?
-
@Dashrender said:
@thecreativeone91 I take it you can't justify making a SSD machine with enough storage to work on?
If you want to pay for 20TB of SSD storage and ship it to me go ahead. I could probably do with 12TB if I had too, and move projects each time I work on them.
-
LOL I was only asking.
-
I have always liked th StoragePod design. I've never gotten to work with one, but I eventually want to buld one.
I like their Vault ideology (https://www.backblaze.com/blog/vault-cloud-storage-architecture/)... It improves availability and reliability over their individual Storage Pods... But I'd like the a bility to start with 2 Pods to make a vault and then scale up, lol.
-
It's interesting but I don't think it makes sense unless it's at a large scale.
-
Agreed, it is only mildly interesting because it is not an enterprise chassis but just a very cheap way of holding a bunch of drives sans hot swap. It's effectively useless for anything in the SMB or even in a home. Too slow, too fragile, too big. In a cluster, as a node, as it is designed to be used it makes perfect sense. Anywhere else, it's so pointless as to be boring and pedestrian. There is no special technology in it, nothing interesting.
It's the SAM-SD without the enterprise bit - which kind of defeats the purpose. For just a little more money you can get an enterprise chassis that is faster, more reliable, has hot swap bays, takes less work and actually makes sense for SMBs or home users with large storage needs.
-
@thecreativeone91 Agreed... But I think it would be easy-ish to start small and add drives to each unit as you go... If you were to start with say 3 pods and 10 x 2TB drives each...
I theorize about stuff like this a lot, lol (thus the reason I want to build a small one and play, lol)... but I lack the personal budget to do it with.
-
@dafyre said:
@thecreativeone91 Agreed... But I think it would be easy-ish to start small and add drives to each unit as you go... If you were to start with say 3 pods and 10 x 2TB drives each...
I theorize about stuff like this a lot, lol (thus the reason I want to build a small one and play, lol)... but I lack the personal budget to do it with.
How is that good, though? Three pods does not a cluster make and is crazy expensive and growing would be essentially impossible at that scale. Using units like these with 2TB drives doesn't make sense, you would use much smaller units with bigger drives for less, and growing storage over time is a huge waste of money and big risk, growing the individual units doesn't work because of the parity rebuild cycles and buying three pods for the capacity of less than one doesn't work financially.
Until you need many PB of data and need it today, these have no place. And only if you have custom storage systems to do the scale out because just having these in a "cluster" is not enough, BackBlaze's secret sauce that makes these actually useful is just that.... secret. The concept isn't secret, but the software is.
-
Three pods with RAIN and 10x2TB in RAID 6 is roughly 16TB usable capacity if well protected and 32TB if loosely protected (and not enough to even try getting to AetherStore level protection!)
That is an insane amount of money to pay for 16TB usable capacity. You could get to a similar capacity with a $600 ReadyNAS four bay unit and modern Helium drives in RAID 10!
-
Food for thought then. I knew it only makes sense at scale. I guess my math was off then. I figured it would make sense at a smaller scale and grow as you go, but I guess not.
At the scale you are talking about, it makes sense for BackBlaze to actually sell their Vault line... Just buy a rack or two of storage from them, lol.
-
@dafyre said:
Food for thought then. I knew it only makes sense at scale. I guess my math was off then. I figured it would make sense at a smaller scale and grow as you go, but I guess not.
At the scale you are talking about, it makes sense for BackBlaze to actually sell their Vault line... Just buy a rack or two of storage from them, lol.
It's scale out, but you would never use a partially populated one. You would probably not consider these at all until you need ten to twenty fully populated units on day one (think in terms of full racks, not full servers here!!) and you would grow by popping in a fully populated unit at a time.
So you would start, at a bare minimum, with something like 2.5PB to start and grew 250TB increments. At no point would you ever plan to do a parity rebuild. You might do it in case of a failure, but you might not.
-
@scottalanmiller said:
Three pods with RAIN
My servers would never forgive me for putting them in the RAIN... lol
