Comment by dylan604
2 days ago
SSD speeds are nothing short of miraculous in my mind. I come from the old days of striping 16 HDDs together (at a minimum number) to get 1GB/s throughput. Depending on the chassis, that was 2 8-drive enclosures in the "desktop" version or the large 4RU enclosures with redundant PSUs and fans loud enough to overpower arena rock concerts. Now, we can get 5+GB/s throughput from a tiny stick that can be used externally via a single cable for data&power that is absolutely silent. I edit 4K+ video as well, and now can edit directly from the same device the camera recorded to during production. I'm skipping over the parts of still making backups, but there's no more multi-hour copy from source media to edit media during a DIT step. I've spent many a shoot as a DIT wishing the 1s&0s would travel across devices much faster while everyone else on the production has already left, so this is much appreciated by me. Oh, and those 16 device units only came close to 4TB around the time of me finally dropping spinning rust.
The first enclosure I ever dealt with was a 7-bay RAID-0 that could just barely handle AVR75 encoding from Avid. Just barely to the point that only video was saved to the array. The audio throughput would put it over the top, so audio was saved to a separate external drive.
Using SSD feels like a well deserved power up from those days.
The latency of modern NVMe is what really blows my mind (as low as 20~30 uS). NVMe is about an order of magnitude quicker than SAS and SATA.
This is why I always recommend developers try using SQLite on top of NVMe storage. The performance is incredible. I don't think you would see query times anywhere near 20uS with a hosted SQL solution, even if it's on the same machine using named pipes or other IPC mechanism.
Meanwhile a job recently told me they are on IBM AS400 boxes “because Postgress and other sql databases can’t keep up with the number of transactions we have”… for a company that has a few thousand inserts per day…
Obviously not true that they’d overwhelm modern databases but feels like that place has had the same opinions since the 1960s.
Then there's optane that got ~10us with. The newest controllers and nand is inching closer with randoms but optane is still the most miraculous ssd tech that's normally obtainable
But they've retired it??
Optane is no longer available :(
3 replies →
Eventually we'll have machines with unified memory+storage. You'll certainly have to take a bit of a performance hit in certain scenarios but also think about the load time improvements. If you store video game files in the same format they'd be needed at runtime you could be at the main menu in under a second.
9 replies →
I’m still buying old Optane drives for that latency when it matters. RockDB loves it
>Optane drives
This is the secret hack a lot of people don't know about!
It's not really the SSDs themselves that are incredibly fast (they still are somewhat), it's mostly the RAM cache and clever tricks to make TLC feel like SLC.
Most (cheap) SSDs their performance goes off a cliff once you hit the boundary of these tricks.
[flagged]
It can be good to know that SSDs are fast until you exhaust the cache by copying gigs of files around.
It doesn’t hurt to be aware of the limitations even if for the common case things are much better.
1 reply →
Not sure what warranted such an aggressive response.
I grew up in the 90s, on 56kb modems and PCs that rumbled and whined when you booted them up. I was at the tail end of using floppies.
I never said I didn't love the speed of SSDs, and when they just started to become mainstream it was an upgrade I did for everyone around me. I made my comment in part because you mentioned dumping 4K into the SSD and/or editing it. It can be a nasty surprise if you're doing something live, and suddenly your throughout plummets, everything starts to stutter and you have no idea why.
2 replies →
> once you hit the boundary of these trick
Tell me more. When do I hit the boundary? What is perf before/after said boundary? What are the tricks?
Tell me something actionable. Educate me.
Your tone is quite odd here. I'm having difficulty parsing your intention, but I'm going to assume you're being genuine because why not.
For the RAM cache, you hit the boundaries when you exhaust the RAM cache. It performs faster, but is smaller and once full, data has to be off/loaded at the rate of the slower backing NAND. It might not be RAM, either, sometimes faster SLC NAND is used for the cache.
It's not really possible to describe it much more concretely than that beyond what you've already been told, performance falls off a cliff when that happens. How long "it" takes, what the level of performance is before and after, it all depends on the device.
There are many more tricks that SSD manufacturers use, but caching is the only one I know of related to speed so I'll leave the rest in the capable hands of Google.
10 replies →
Toms Hardware usually includes a "Sustained Write Performance and Cache Recovery" test
The test measures the write cache speed and the time to the fall to the native NAND write speed. There are usually irregularities in the sustained write speeds as well.
https://www.tomshardware.com/reviews/wd-black-sn850x-ssd-rev...
The other test I've seen is based on writing and using up free space, SSD performance can drop off as the free space fills up and garbage collection efficiency goes down. I think this impacts random writes particularly
In the enterprise space, drives tend to keep more over provisioned NAND free to maintain more consistent performance. Very early on the SSD timeline, it was advisable to only allocate 80% of consumer drives if you were using them outside of desktops and expected the workload to fill them.
This hits home even more since I started restoring some vintage Macs.
For the ones new enough to get an SSD upgrade, it's night and day the difference (even a Power Mac G4 can feel fresh and fast just swapping out the drive). For older Macs like PowerBooks and classic Macs, there are so many SD/CF card to IDE/SCSI/etc. adapters now, they also get a significant boost.
But part of the nostalgia of sitting there listening to the rumble of the little hard drive is gone.
> But part of the nostalgia of sitting there listening to the rumble of the little hard drive is gone.
I remember this being a key troubleshooting step. Listen/feel for the hum of the hard drive OR the telltale click clack, grinding, etc that foretold doom.
Thank the gawds we no longer have to worry about the click of death
1 reply →
I've just finished CF swapping a PowerBook 1400cs/117. It's a base model with 12MB RAM, so there are other bottlenecks, but OS 8.1 takes about 90 seconds from power to desktop and that's pretty good for a low-end machine with a fairly heavy OS.
Somehow the 750MB HDD from 1996 is still working, but I admit that the crunch and rumble of HDDs is a nostalgia I'm happy to leave in the past.
My 1.67 PowerBook G4 screams with a 256GB mSATA SSD-IDE adapter. Until you start compiling code or web surfing, it still feels like a pretty modern machine. I kind of wish I didn't try the same upgrade on a iBook G3, though...
>I kind of wish I didn't try the same upgrade on a iBook G3, though...
Oh god. Those were the worst things ever to upgrade the hard drive. Just reading this gave me a nightmare flashback to having to keep track of all the different screws. This is why my vintage G3 machine is a Pismo instead of an iBook.
1 reply →
I just picked up a 1.5GHz Powerbook G4 12-inch in mint condition. RAM is maxed out but I've been putting off the SSD-IDE upgrade because of how intrusive it is and many screws are involved.
I had a 2011 MBP that I kept running by replacing the HDD with an SSD, and then removed the DVD-ROM drive with a second SSD. The second SSD had throughput limits because it was designed for shiny round disc, so it had a lower ability chip. I had that until the 3rd GPU replacement died, and eventually switched to second gen butterfly keyboard. The only reason it was tolerable was because of the SSDs, oh and the RAM upgrades
Did you ever have the GPU issue? My sister had a 2011, I had to desolder a resistor (or maybe two?) on it to bypass the dGPU since it was causing it to boot loop. But now it's still running and pretty happily for some basic needs!
3 replies →
While these are geared toward retrocomputing, there are things that attempt to simulate the sound based on activity LEDs: https://www.serdashop.com/HDDClicker
> For older Macs like PowerBooks and classic Macs, there are so many SD/CF card to IDE/SCSI/etc.
Would those be bandwidth limited by the adapter/card or CPU? Can you get throughput higher than say, a cheap 2.5" SSD over Sata 3/4?
You are limited at first by the IDE/SCSI interface, so below SATA speeds.
1 reply →
You should try now-discontinued Intel Optane especially p5800x. I got my OS running on them and they are incredible.
I'm running 12 of them for ZFS cache/log/special, and they are fast/tough enough to make a large array on a slow link feel fast. I shake my fist at Intel and Micron for taking away one of the best memory technologies to ever exist.
The endurance on those drivers is amazing.
I have (stupidly) used a too small Samsung EVO drive as a caching drive, and that is probably the first computer part that I've worn out (bar a mouse & keyboard).
Just a few more years until we get MRAM as viable storage technology. And affordable fusion, and hovercars.
Totally. I spent a lot of time 15-20 years ago building out large email systems.
I recently bought a $17 SSD for my son’s middle school project that was speced to deliver like 3x what I needed in those days. From a storage perspective, I was probably spending $50 GB/mo all-in to deploy a multi million dollar storage solution. TBH… you’d probably smoke that system with used laptops today.
If only folks wrote software with this in mind.
> I come from the old days of striping 16 HDDs together (at a minimum number) to get 1GB/s throughput
Woah, how long would that last before you'd start having to replace the drives?
If you're interested in some hard data, Backblaze publishes their HD failure numbers[1]. These disks are storage optimized, not performance optimized like the parent comment, but they have a pretty large collection of various hard drives, and it's pretty interesting to see how reliability can vary dramatically across brand and model.
---
1. https://www.backblaze.com/cloud-storage/resources/hard-drive...
The Backblaze reports are impressive. It would have been very handy to know which models to buy. They break it down to capacity of the same family of drives so a 2TB might be sound, but the 4TB might be more flaky. That information is very useful when it comes time to think about upgrading capacity in the arrays. Having someone go through these battles and then give away the data learned would just be dumb to not take advantage of their generosity.
2 replies →
Depending on the HDD vendor/model. We had hot spares and cold spares. On one build, we had a bad batch of drives. We built the array on a Friday, and left it for burn-in running over the weekend. On Monday, we came in to a bunch of alarms and >50% failure rate. At least they died during the burn-in so no data loss, but it was an extreme example. That was across multiple 16-bay rack mount chassis. It was an infamous case though, we were not alone.
More typically, you'd have a drive die much less frequently, but it was something you absolutely had to be prepared for. With RAID-6 and a hot spare, you could be okay with a single drive failure. Theoretically, you could lose two, but it would be a very nervy day getting the array to rebuild without issue.
I asked because I did a makeshift NAS for myself with three 4tb ironwolf, but they died before the third year. I didn't investigate much, but it was most likely because of power outages and a lack of a nobreak PSU at that time. It's still quite a bit of work to maintain physical hard drives and the probability of failure as I understand tend to increase the more units the array has because of inverse probability (not the likelihood of one of them failing but the likelihood of none of them failing after a period of time, which is cumulative)
1 reply →
I run 24x RAID at home. I’m replacing disks 2-3 times per year.
Are your drives under heavy load or primarily just spinning waiting for use? Are they dying unsuspectedly, or are you watching the SMART messages and being prepared when it happens?
3 replies →
In a similar vein, I had Western Digital Raptors striped in my gaming PC in the mid 2000s. I remember just how amazed I was after moving to SSDs.
So, now someone can strip several SSDs to gain more performance as well for other purposes than video editing, right?
[dead]