search

RemixVSL / iomemory-vsl

Updated Fusion-io iomemory VSL Linux (version 3.2.16) driver for recent kernels.
153 stars 28 forks source link

vsl drivers for Linux 5.16? #99

Closed 4nanook closed 2 years ago

4nanook commented 2 years ago 
  I am trying to compile drivers for 5.16 not having luck with the existing package, says can't find stddef.h even though it is in the headers tree.  I just grabbed the source using git clone and did not commit any particular release as this was the instructions I was given for 5.13, because all version of Linux prior to 5.16 have a bug that can result in file corruption and/or privilege escalation, it is desirable to be on the latest kernel, 5.16.14, which has addressed these issues.  Previously I was on 5.13.19 and it compiled okay under that kernel. Kernels are self-compiled, header files get stuck in /usr/src/linux-headers-5.13.19 or /usr/src/linux-headers-5.16.14.  Thanks in advance for any help you can offer.
bplein commented 2 years ago 
  I am trying to compile drivers for 5.16 not having luck with the existing package, says can't find stddef.h even though it is in the headers tree.  I just grabbed the source using git clone and did not commit any particular release as this was the instructions I was given for 5.13, because all version of Linux prior to 5.16 have a bug that can result in file corruption and/or privilege escalation, it is desirable to be on the latest kernel, 5.16.14, which has addressed these issues.  Previously I was on 5.13.19 and it compiled okay under that kernel. Kernels are self-compiled, header files get stuck in /usr/src/linux-headers-5.13.19 or /usr/src/linux-headers-5.16.14.  Thanks in advance for any help you can offer.

Significant Linux kernel changes in 5.15 have broken this driver for now. The primary developer will eventually work on it but it's broken for now. Suggest going back to 5.14 or newer if you can.

4nanook commented 2 years ago 
  Thanks, but because I need my machines to be secure and 5.14 does not

provide this and the exploit is now out in the public, I'm going to replace this drive with a western digital drive that is supported by Linux kernel out of the box. I will lose the atomic write capability but the database is capable of 10,000 TPS with this drive, where it maxed at about 200 TPS with rotory drives, however when it maxes at 10,000 the machine is maxed on CPU at that point and realistically the load normally does not exceed about 400 TPS so I think I'll be fine with the WD.

---------------------------------------_- Eskimo North Linux Friendly Internet Access, Shell Accounts, and Hosting. Knowledgeable human assistance, not telephone trees or script readers. See our web site: http://www.eskimo.com/ (206) 812-0051 or (800) 246-6874.

On Sat, 12 Mar 2022, Bill Plein wrote:

Date: Sat, 12 Mar 2022 07:23:00 -0800 From: Bill Plein @.> Reply-To: RemixVSL/iomemory-vsl @.> To: RemixVSL/iomemory-vsl @.> Cc: Robert Dinse @.>, Author @.***> Subject: Re: [RemixVSL/iomemory-vsl] vsl drivers for Linux 5.16? (Issue #99)

  I am trying to compile drivers for 5.16 not having luck with the existing package, says can't find stddef.h even though it is in the headers tree.  I just grabbed the source using git clone and did not commit any particular release as this was the instructions I was given for 5.13, because all version of Linux prior to 5.16 have a bug that can result in file corruption and/or privilege escalation, it is desirable to be on the latest kernel, 5.16.14, which has addressed these issues.  Previously I was on 5.13.19 and it compiled okay under that kernel. Kernels are self-compiled, header files get stuck in /usr/src/linux-headers-5.13.19 or /usr/src/linux-headers-5.16.14.  Thanks in advance for any help you can offer.

Significant Linux kernel changes in 5.15 have broken this driver for now. The primary developer will eventually work on it but it's broken for now. Suggest going back to 5.14 or newer if you can.

-- Reply to this email directly or view it on GitHub: https://github.com/RemixVSL/iomemory-vsl/issues/99#issuecomment-1065901474 You are receiving this because you authored the thread.

Message ID: @.***>

sakurachan00 commented 2 years ago

Thanks, but because I need my machines to be secure and 5.14 does not provide this and the exploit is now out in the public, I'm going to replace this drive with a western digital drive that is supported by Linux kernel out of the box. I will lose the atomic write capability but the database is capable of 10,000 TPS with this drive, where it maxed at about 200 TPS with rotory drives, however when it maxes at 10,000 the machine is maxed on CPU at that point and realistically the load normally does not exceed about 400 TPS so I think I'll be fine with the WD. … ---------------------------------------_- Eskimo North Linux Friendly Internet Access, Shell Accounts, and Hosting. Knowledgeable human assistance, not telephone trees or script readers. See our web site: http://www.eskimo.com/ (206) 812-0051 or (800) 246-6874. On Sat, 12 Mar 2022, Bill Plein wrote: Date: Sat, 12 Mar 2022 07:23:00 -0800 From: Bill Plein @.> Reply-To: RemixVSL/iomemory-vsl @.> To: RemixVSL/iomemory-vsl @.> Cc: Robert Dinse @.>, Author @.> Subject: Re: [RemixVSL/iomemory-vsl] vsl drivers for Linux 5.16? (Issue #99) > I am trying to compile drivers for 5.16 not having luck with the existing package, says can't find stddef.h even though it is in the headers tree. I just grabbed the source using git clone and did not commit any particular release as this was the instructions I was given for 5.13, because all version of Linux prior to 5.16 have a bug that can result in file corruption and/or privilege escalation, it is desirable to be on the latest kernel, 5.16.14, which has addressed these issues. Previously I was on 5.13.19 and it compiled okay under that kernel. Kernels are self-compiled, header files get stuck in /usr/src/linux-headers-5.13.19 or /usr/src/linux-headers-5.16.14. Thanks in advance for any help you can offer. Significant Linux kernel changes in 5.15 have broken this driver for now. The primary developer will eventually work on it but it's broken for now. Suggest going back to 5.14 or newer if you can. -- Reply to this email directly or view it on GitHub: #99 (comment) You are receiving this because you authored the thread. Message ID: @.>

I'd use LTS kernel branches as they'd be patched for stuff like this

bplein commented 2 years ago

@4nanook Is this for a production environment in a work environment? These cards are a decade old, and that should be taken into consideration in production environments as well.

Even at home I have moved my primary data to NVMe U.2 drives because I know a kernel change won't render NVMe drives useless, and use the Fusion-io cards instead for high performance temporary workspace and backups.

4nanook commented 2 years ago 
  Unfortunately the kernels chosen for LTS releases have all provided crap

performance for my applications. I'm really stretching the hardware to the limit which is why I'm bothering with self-compiled kernels to begin with. 5.13 provided good performance, but not LTS, 5.14 and 5.15 both spent more time in system space than user and did not work at all well, 5.15 was the most recent LTS, 5.16 again good performance. And so it goes.

---------------------------------------_- Eskimo North Linux Friendly Internet Access, Shell Accounts, and Hosting. Knowledgeable human assistance, not telephone trees or script readers. See our web site: http://www.eskimo.com/ (206) 812-0051 or (800) 246-6874.

On Sun, 13 Mar 2022, sakurachan00 wrote:

Date: Sun, 13 Mar 2022 06:33:46 -0700 From: sakurachan00 @.> Reply-To: RemixVSL/iomemory-vsl @.> To: RemixVSL/iomemory-vsl @.> Cc: Robert Dinse @.>, Author @.***> Subject: Re: [RemixVSL/iomemory-vsl] vsl drivers for Linux 5.16? (Issue #99)

Thanks, but because I need my machines to be secure and 5.14 does not provide this and the exploit is now out in the public, I'm going to replace this drive with a western digital drive that is supported by Linux kernel out of the box. I will lose the atomic write capability but the database is capable of 10,000 TPS with this drive, where it maxed at about 200 TPS with rotory drives, however when it maxes at 10,000 the machine is maxed on CPU at that point and realistically the load normally does not exceed about 400 TPS so I think I'll be fine with the WD. … ---------------------------------------_- Eskimo North Linux Friendly Internet Access, Shell Accounts, and Hosting. Knowledgeable human assistance, not telephone trees or script readers. See our web site: http://www.eskimo.com/ (206) 812-0051 or (800) 246-6874. On Sat, 12 Mar 2022, Bill Plein wrote: Date: Sat, 12 Mar 2022 07:23:00 -0800 From: Bill Plein @.> Reply-To: RemixVSL/iomemory-vsl @.> To: RemixVSL/iomemory-vsl @.> Cc: Robert Dinse @.>, Author @.> Subject: Re: [RemixVSL/iomemory-vsl] vsl drivers for Linux 5.16? (Issue #99) > I am trying to compile drivers for 5.16 not having luck with the existing package, says can't find stddef.h even though it is in the headers tree. I just grabbed the source using git clone and did not commit any particular release as this was the instructions I was given for 5.13, because all version of Linux prior to 5.16 have a bug that can result in file corruption and/or privilege escalation, it is desirable to be on the latest kernel, 5.16.14, which has addressed these issues. Previously I was on 5.13.19 and it compiled okay under that kernel. Kernels are self-compiled, header files get stuck in /usr/src/linux-headers-5.13.19 or /usr/src/linux-headers-5.16.14. Thanks in advance for any help you can offer. Significant Linux kernel changes in 5.15 have broken this driver for now. The primary developer will eventually work on it but it's broken for now. Suggest going back to 5.14 or newer if you can. -- Reply to this email directly or view it on GitHub: #99 (comment) You are receiving this because you authored the thread. Message ID: @.>

I'd use LTS kernel branches as they'd be patched for stuff like this

-- Reply to this email directly or view it on GitHub: https://github.com/RemixVSL/iomemory-vsl/issues/99#issuecomment-1066103512 You are receiving this because you authored the thread.

Message ID: @.***>

4nanook commented 2 years ago 
  Yes I realize that.  The reason I went with these cards was specifically

a recommendation of the MariaDB engineers. And it did get me a database that handled more than 10,000 TPS, so can't complain where on rotory drives I was maxing out at a little more than 200 TPS, but I think even a modern regular flash drive will get me where I need to be on database performance so thats what I'm going to attempt.

---------------------------------------_- Eskimo North Linux Friendly Internet Access, Shell Accounts, and Hosting. Knowledgeable human assistance, not telephone trees or script readers. See our web site: http://www.eskimo.com/ (206) 812-0051 or (800) 246-6874.

On Sun, 13 Mar 2022, Bill Plein wrote:

Date: Sun, 13 Mar 2022 10:48:45 -0700 From: Bill Plein @.> Reply-To: RemixVSL/iomemory-vsl @.> To: RemixVSL/iomemory-vsl @.> Cc: Robert Dinse @.>, Mention @.***> Subject: Re: [RemixVSL/iomemory-vsl] vsl drivers for Linux 5.16? (Issue #99)

@4nanook Is this for a production environment in a work environment? These cards are a decade old, and that should be taken into consideration in production environments as well.

Even at home I have moved my primary data to NVMe U.2 drives because I know a kernel change won't render them useless, and use them instead for high performance temporary workspace and backups.

-- Reply to this email directly or view it on GitHub: https://github.com/RemixVSL/iomemory-vsl/issues/99#issuecomment-1066149978 You are receiving this because you were mentioned.

Message ID: @.***>

Tourman36 commented 2 years ago

Honestly while these cards were good and performance can be good, we are limited by only having access to the kernel source code.

I just at this point wouldn’t recommend these in any sort of critical infrastructure. You risk the potential from data loss due to unforeseen complications with the hw driver and kernel. You can poke around in our discord channel and talk to us there.

You would really be better served by using NVME drives which are natively supported in all 5.X kernels. They can be had for relatively cheap these days and should offer similar if not better performance.

On Sun, Mar 13, 2022 at 5:32 PM Robert Dinse @.***> wrote:

Yes I realize that. The reason I went with these cards was specifically a recommendation of the MariaDB engineers. And it did get me a database that handled more than 10,000 TPS, so can't complain where on rotory drives I was maxing out at a little more than 200 TPS, but I think even a modern regular flash drive will get me where I need to be on database performance so thats what I'm going to attempt.

---------------------------------------_- Eskimo North Linux Friendly Internet Access, Shell Accounts, and Hosting. Knowledgeable human assistance, not telephone trees or script readers. See our web site: http://www.eskimo.com/ (206) 812-0051 or (800) 246-6874.

On Sun, 13 Mar 2022, Bill Plein wrote:

Date: Sun, 13 Mar 2022 10:48:45 -0700 From: Bill Plein @.> Reply-To: RemixVSL/iomemory-vsl @.> To: RemixVSL/iomemory-vsl @.> Cc: Robert Dinse @.>, Mention @.***> Subject: Re: [RemixVSL/iomemory-vsl] vsl drivers for Linux 5.16? (Issue

99)

@4nanook Is this for a production environment in a work environment? These cards are a decade old, and that should be taken into consideration in production environments as well.

Even at home I have moved my primary data to NVMe U.2 drives because I know a kernel change won't render them useless, and use them instead for high performance temporary workspace and backups.

-- Reply to this email directly or view it on GitHub:

https://github.com/RemixVSL/iomemory-vsl/issues/99#issuecomment-1066149978 You are receiving this because you were mentioned.

Message ID: @.***>

— Reply to this email directly, view it on GitHub https://github.com/RemixVSL/iomemory-vsl/issues/99#issuecomment-1066186284, or unsubscribe https://github.com/notifications/unsubscribe-auth/AO7IG5G4OXDNR2UHPPMXBD3U7ZNHJANCNFSM5QRV7RHQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you are subscribed to this thread.Message ID: @.***>

dbartelmus commented 2 years ago

So consumer NVMe will be better than IO Drive 2 nowadays? IOPS are ok but what about access latency and small file throughput?

4nanook commented 2 years ago 
  The NVMe drives don't offer atomic writes which was why the engineer at

MariaDB recommended the Fusion I/O, however, it turned out atomic writes were not supported by the ext4 file system I was using anyway so never utilized that feature and still achieved 10,000 TPS, so I've bought WD Black 750's to replace the I/O fusion drives, been using one on my workstation for several years and so far no errors and smartctl still indicates another three years projected lifetime, so that's where I'm going for now. And this is all just a stop-gap while I build a newer server with quite a lot more resources. This particular server does many things but the largest portion of the load on the database is a friendica social media site. It is part of the fediverse and so handles a large message volume. We make daily backups to a RAIDED rotary drive file system so worst that will happen is we'll lose a day of messages and/or wordpress updates. The Linux folks have been tweaking the schedular and the block I/O code quite a lot and both of those things can very heavily impact the peformance, and the existing machine is a six core box running at 4.2 Ghz, not enough, so building a 14-core 28-thread machine with twice the RAM that I hope will give us a bit more overhead.

---------------------------------------_- Eskimo North Linux Friendly Internet Access, Shell Accounts, and Hosting. Knowledgeable human assistance, not telephone trees or script readers. See our web site: http://www.eskimo.com/ (206) 812-0051 or (800) 246-6874.

On Sun, 13 Mar 2022, Demetrius Cassidy wrote:

Date: Sun, 13 Mar 2022 14:46:51 -0700 From: Demetrius Cassidy @.> Reply-To: RemixVSL/iomemory-vsl @.> To: RemixVSL/iomemory-vsl @.> Cc: Robert Dinse @.>, Mention @.***> Subject: Re: [RemixVSL/iomemory-vsl] vsl drivers for Linux 5.16? (Issue #99)

Honestly while these cards were good and performance can be good, we are limited by only having access to the kernel source code.

I just at this point wouldn’t recommend these in any sort of critical infrastructure. You risk the potential from data loss due to unforeseen complications with the hw driver and kernel. You can poke around in our discord channel and talk to us there.

You would really be better served by using NVME drives which are natively supported in all 5.X kernels. They can be had for relatively cheap these days and should offer similar if not better performance.

On Sun, Mar 13, 2022 at 5:32 PM Robert Dinse @.***> wrote:

Yes I realize that. The reason I went with these cards was specifically a recommendation of the MariaDB engineers. And it did get me a database that handled more than 10,000 TPS, so can't complain where on rotory drives I was maxing out at a little more than 200 TPS, but I think even a modern regular flash drive will get me where I need to be on database performance so thats what I'm going to attempt.

---------------------------------------_- Eskimo North Linux Friendly Internet Access, Shell Accounts, and Hosting. Knowledgeable human assistance, not telephone trees or script readers. See our web site: http://www.eskimo.com/ (206) 812-0051 or (800) 246-6874.

On Sun, 13 Mar 2022, Bill Plein wrote:

Date: Sun, 13 Mar 2022 10:48:45 -0700 From: Bill Plein @.> Reply-To: RemixVSL/iomemory-vsl @.> To: RemixVSL/iomemory-vsl @.> Cc: Robert Dinse @.>, Mention @.***> Subject: Re: [RemixVSL/iomemory-vsl] vsl drivers for Linux 5.16? (Issue

99)

@4nanook Is this for a production environment in a work environment? These cards are a decade old, and that should be taken into consideration in production environments as well.

Even at home I have moved my primary data to NVMe U.2 drives because I know a kernel change won't render them useless, and use them instead for high performance temporary workspace and backups.

-- Reply to this email directly or view it on GitHub:

https://github.com/RemixVSL/iomemory-vsl/issues/99#issuecomment-1066149978 You are receiving this because you were mentioned.

Message ID: @.***>

— Reply to this email directly, view it on GitHub https://github.com/RemixVSL/iomemory-vsl/issues/99#issuecomment-1066186284, or unsubscribe https://github.com/notifications/unsubscribe-auth/AO7IG5G4OXDNR2UHPPMXBD3U7ZNHJANCNFSM5QRV7RHQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you are subscribed to this thread.Message ID: @.***>

-- Reply to this email directly or view it on GitHub: https://github.com/RemixVSL/iomemory-vsl/issues/99#issuecomment-1066188343 You are receiving this because you were mentioned.

Message ID: @.***>

4nanook commented 2 years ago 
  It probably is not going to perform as well as the IO-Fusion drive but

presently the Friendica load averages around 400 TPS, with rotary I was only about to obtain about 250 TPS, so if the WD does even half that we are fine, and the 10,000 TPS was CPU maxing out not the drive. The majority of the transactions are reads but seek performance on rotarys made even reads slow. Because this machine is very much in the public eye it gets a lot more hacker attention than I would like so security is particularly important. One of the reasons I'm putting a new machine together aside from more cores is more RAM (it will have 256GB as opposed to 128GB of the existing machine) and more PCI lanes so more I/O bandwidth. More memory will enable it to cache more of the data and more I/O will maximize the ability to utilize whatever hardware it has attached.

---------------------------------------_- Eskimo North Linux Friendly Internet Access, Shell Accounts, and Hosting. Knowledgeable human assistance, not telephone trees or script readers. See our web site: http://www.eskimo.com/ (206) 812-0051 or (800) 246-6874.

On Sun, 13 Mar 2022, Damian Bartelmus wrote:

Date: Sun, 13 Mar 2022 14:58:42 -0700 From: Damian Bartelmus @.> Reply-To: RemixVSL/iomemory-vsl @.> To: RemixVSL/iomemory-vsl @.> Cc: Robert Dinse @.>, Mention @.***> Subject: Re: [RemixVSL/iomemory-vsl] vsl drivers for Linux 5.16? (Issue #99)

So consumer NVMe will be better than IO Drive 2 nowadays? IOPS are ok but what about access latency and small file throughput?

-- Reply to this email directly or view it on GitHub: https://github.com/RemixVSL/iomemory-vsl/issues/99#issuecomment-1066190068 You are receiving this because you were mentioned.

Message ID: @.***>

bplein commented 2 years ago

So consumer NVMe will be better than IO Drive 2 nowadays? IOPS are ok but what about access latency and small file throughput?

I don't use consumer NVMe drives as a replacement for Fusion-io. I use Enterprise/Datacenter devices. https://ark.intel.com/content/www/us/en/ark/products/122579/intel-ssd-dc-p4510-series-4-0tb-2-5in-pcie-3-1-x4-3d2-tlc.html

4nanook commented 2 years ago 
  That's great but the app doesn't justify a $5000 device.

---------------------------------------_- Eskimo North Linux Friendly Internet Access, Shell Accounts, and Hosting. Knowledgeable human assistance, not telephone trees or script readers. See our web site: http://www.eskimo.com/ (206) 812-0051 or (800) 246-6874.

On Mon, 14 Mar 2022, Bill Plein wrote:

Date: Mon, 14 Mar 2022 06:48:17 -0700 From: Bill Plein @.> Reply-To: RemixVSL/iomemory-vsl @.> To: RemixVSL/iomemory-vsl @.> Cc: Robert Dinse @.>, Mention @.***> Subject: Re: [RemixVSL/iomemory-vsl] vsl drivers for Linux 5.16? (Issue #99)

So consumer NVMe will be better than IO Drive 2 nowadays? IOPS are ok but what about access latency and small file throughput?

I don't use consumer NVMe drives as a replacement for Fusion-io. I use Enterprise/Datacenter devices. https://ark.intel.com/content/www/us/en/ark/products/122579/intel-ssd-dc-p4510-series-4-0tb-2-5in-pcie-3-1-x4-3d2-tlc.html

-- Reply to this email directly or view it on GitHub: https://github.com/RemixVSL/iomemory-vsl/issues/99#issuecomment-1066816028 You are receiving this because you were mentioned.

Message ID: @.***>

bplein commented 2 years ago

4TB NVMe U.2 drives are under $400 used.

I'm going to close this issue and re-open one without the chatter.

bplein commented 2 years ago

@4nanook Take a look at https://github.com/RemixVSL/iomemory-vsl/issues/100 and you can try out the code on main

4nanook commented 2 years ago 
  I've ended up extracting the card from the machine as it conflicted with

a network card that I wanted to use to fix a bug in hardware offloading so at the moment do not have a platform with which to participate.

---------------------------------------_- Eskimo North Linux Friendly Internet Access, Shell Accounts, and Hosting. Knowledgeable human assistance, not telephone trees or script readers. See our web site: http://www.eskimo.com/ (206) 812-0051 or (800) 246-6874.

On Tue, 15 Mar 2022, Bill Plein wrote:

Date: Tue, 15 Mar 2022 08:15:09 -0700 From: Bill Plein @.> Reply-To: RemixVSL/iomemory-vsl @.> To: RemixVSL/iomemory-vsl @.> Cc: Robert Dinse @.>, Mention @.***> Subject: Re: [RemixVSL/iomemory-vsl] vsl drivers for Linux 5.16? (Issue #99)

@4nanook Take a look at https://github.com/RemixVSL/iomemory-vsl/issues/100 and you can try out the code on main

-- Reply to this email directly or view it on GitHub: https://github.com/RemixVSL/iomemory-vsl/issues/99#issuecomment-1068106460 You are receiving this because you were mentioned.

Message ID: @.***>

4nanook commented 2 years ago 
  I will give it a try but not until I get the new flash in place.  I

can't boot the machine into the new kernel without interrupting service and I don't wish to do that more than I need to so going to get things moved, on the new kernel, then can try the drivers again.

---------------------------------------_- Eskimo North Linux Friendly Internet Access, Shell Accounts, and Hosting. Knowledgeable human assistance, not telephone trees or script readers. See our web site: http://www.eskimo.com/ (206) 812-0051 or (800) 246-6874.

On Tue, 15 Mar 2022, Bill Plein wrote:

Date: Tue, 15 Mar 2022 08:15:09 -0700 From: Bill Plein @.> Reply-To: RemixVSL/iomemory-vsl @.> To: RemixVSL/iomemory-vsl @.> Cc: Robert Dinse @.>, Mention @.***> Subject: Re: [RemixVSL/iomemory-vsl] vsl drivers for Linux 5.16? (Issue #99)

@4nanook Take a look at https://github.com/RemixVSL/iomemory-vsl/issues/100 and you can try out the code on main

-- Reply to this email directly or view it on GitHub: https://github.com/RemixVSL/iomemory-vsl/issues/99#issuecomment-1068106460 You are receiving this because you were mentioned.

Message ID: @.***>

Tourman36 commented 2 years ago

I did not say consumer nvme. Enterprise grade sure - stay away from consumer gear. Many of them have a low WPD amount and would not survive if you need lots of small writes.

If budget is an issue, I’d go scour eBay. Lots of good second hand gear. Make sure you know what the drive health is before purchasing.

IoDrive is really good for the price point, but is it relatively old technology. We are basically enthusiasts trying to make it functional.

I believe we will still happily accept changes if anyone wants to fix compatibility, but my points remain.

On Sun, Mar 13, 2022 at 5:58 PM Damian Bartelmus @.***> wrote:

So consumer NVMe will be better than IO Drive 2 nowadays? IOPS are ok but what about access latency and small file throughput?

— Reply to this email directly, view it on GitHub https://github.com/RemixVSL/iomemory-vsl/issues/99#issuecomment-1066190068, or unsubscribe https://github.com/notifications/unsubscribe-auth/AO7IG5EVUB3HMMPXG26I5PTU7ZQJFANCNFSM5QRV7RHQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

You are receiving this because you commented.Message ID: @.***>