|
I mean Nimble does have their expandable hybrid-flash arrays that last I looked were somewhat affordable.
|
# ? Aug 19, 2021 03:01 |
|
|
# ? May 11, 2024 15:25 |
|
We use about 20 TB SSD and 50 TB HDD in our datacenters, block only. I have a Dell ME series unit. It, uh, gets the job done, I suppose. I'm not familiar with any other manufacturers units that offer a 10 GbE iSCSI chassis that will take disks and flash, and there's neither money nor desire to run all flash. We just don't need it.
|
# ? Aug 19, 2021 06:36 |
|
In our latest Netapp purchase we got 150TB of SSD and 6PB of NL-SAS. I don't think Pure is our choice in the near future. At least this tender round was won by Netapp and it fit easily within our existing infrastructure. Previous tender was won by EMC and we decided not to purchase anything, we had just gotten rid of our last piece of EMC.
|
# ? Aug 19, 2021 10:52 |
|
idk how the vendor agreements shake out at the single-array scale for Pure but +1 for them. Everything just worked.
|
# ? Aug 19, 2021 12:49 |
|
I picked up a white labelled dell r740xd recently for a cheap price. It came with 256gb ssd for OS? And 1x 8tb SAS drive. I think its got a perc740 RAID card. Currently it's running ESXi. I should be able to buy a bunch of cheap drive sleds off Amazon and load it up with cheap SATA drives right?
|
# ? Aug 20, 2021 05:20 |
|
What are the downsides of top-loading storage chassis vs front/rear in traditional chassis design? The top-loaders offer more disk bays per dollar, and take up less rack space per disk bay. Specifically, I'm comparing things like 36-bay front/rear capacity storage-focused traditional chassis vs 45 or 60-bay top loaders, which seem to offer a better value. Are disk temperatures going to be significantly higher on the top loaders?
|
# ? Sep 24, 2021 18:59 |
|
Twerk from Home posted:What are the downsides of top-loading storage chassis vs front/rear in traditional chassis design? The top-loaders offer more disk bays per dollar, and take up less rack space per disk bay. I'd say the biggest issue with top-load is that you have to pull the chassis out of the rack to replace a HDD. Yes I'm sure that if it is properly installed with the right rails and a suitable cable arm fitted you could do the procedure without powering-off the chassis. However compared to front-load the chance of something going wrong is much higher. Regarding disk temperatures one would assume that a higher density would lead to higher temperatures however it's really hard to say for sure as it depends on the layout, air-flow and workload.
|
# ? Sep 24, 2021 19:44 |
|
I just received an IBM FlashSystem 5035. After configuring the IP in the network setup, I change the cable to Port 1 and change the IP on my device to be in the same range as the IP of the SAN - However, the NIC ports doesn't show any link or activity lights. I can see the NIC adapters through the Technician UI but no dice getting to the initial configuration one the non-Technician ports. I got a service contract on it, but I figured I would see if any goons had any experience with this before going through the hassle. Update: I figured it out. Apparently Node1 Technician port turns into the 'Tech port' and Technician port on Node2 turns into the Initial Configuration - Despite IBMs documentation saying that Port 1 on same node turns into the Initial Config port. Welp. Happy Dolphin fucked around with this message at 14:06 on Oct 4, 2021 |
# ? Oct 4, 2021 13:11 |
Pile Of Garbage posted:I'd say the biggest issue with top-load is that you have to pull the chassis out of the rack to replace a HDD. Yes I'm sure that if it is properly installed with the right rails and a suitable cable arm fitted you could do the procedure without powering-off the chassis. However compared to front-load the chance of something going wrong is much higher.
|
|
# ? Oct 4, 2021 13:50 |
|
BlankSystemDaemon posted:There's still the issue of gyroscopic precession causing wear and tear on the platters and motors when you're moving up to 90 disks at a time. Yeah true, didn't even consider that. Definitely only an issue with top-load where you have to move the chassis to replace disks.
|
# ? Oct 4, 2021 14:22 |
|
Weeks of back-and-forth with Pure support before we got a tech who pointed out that after we create our VIFs and assign them IP addresses, we have to assign them IP addresses. Please tell me this isn't normal support, and that Purity//FA isn't always this dumb?
|
# ? Apr 14, 2022 14:47 |
|
We're using FC, so I can't speak to that side of things, but overall I have been quite happy with our Pure arrays. My only real gripe is that you can't do updates yourself, which seems weird to me in 2022.
|
# ? Apr 14, 2022 16:10 |
|
Internet Explorer posted:We're using FC, so I can't speak to that side of things, but overall I have been quite happy with our Pure arrays. My only real gripe is that you can't do updates yourself, which seems weird to me in 2022. Our FA shipped with 6.1.13 on it - that explains why there's no way to flash it to something newer. Does Pure reach out to you to coordinate an update?
|
# ? Apr 14, 2022 16:22 |
|
Internet Explorer posted:We're using FC, so I can't speak to that side of things, but overall I have been quite happy with our Pure arrays. My only real gripe is that you can't do updates yourself, which seems weird to me in 2022. Hell no. I don’t manage our pures anymore, but when I did it pretty much was “call support, schedule the upgrade, put in CR, enable support access in the right window” and sleep like a baby. When we did our evergreen controller upgrades, it was middle of the day Monday with zero issues.
|
# ? Apr 14, 2022 16:27 |
|
Mierdaan posted:Our FA shipped with 6.1.13 on it - that explains why there's no way to flash it to something newer. Does Pure reach out to you to coordinate an update? You schedule in Pure1. devmd01 posted:Hell no. I don’t manage our pures anymore, but when I did it pretty much was “call support, schedule the upgrade, put in CR, enable support access in the right window” and sleep like a baby. I dunno. Controller upgrade should be bulletproof. If it's bulletproof, let me just press a button or schedule it myself.
|
# ? Apr 14, 2022 16:41 |
|
devmd01 posted:Hell no. I don’t manage our pures anymore, but when I did it pretty much was “call support, schedule the upgrade, put in CR, enable support access in the right window” and sleep like a baby. Nimble I used to rip software upgrades whenever (mid workday included). I have a handful of Unity XT arrays now, they keep insisting that I have a support ticket open before doing anything. Unsure how I feel about that, but the performance and capacity has been untouched in terms of value so far. Also zero issues (knocks on wood).
|
# ? Apr 14, 2022 21:08 |
|
Apparently the Pure gods have heard my prayers. They have added self-service updates in Purity 6.3.
|
# ? May 5, 2022 15:54 |
|
Internet Explorer posted:Apparently the Pure gods have heard my prayers. They have added self-service updates in Purity 6.3. Just in time for Pure stuff to hit the hobbyist market
|
# ? May 5, 2022 16:30 |
|
I work with something like 12 pure boxes and one flashblade, I swear if I could set fire to the flashblade and not get sent straight to the poor house, I'd consider it.
|
# ? Dec 14, 2022 13:10 |
|
Internet Explorer posted:Apparently the Pure gods have heard my prayers. They have added self-service updates in Purity 6.3. I was gonna say, refusing self-service updates just sounds like a way to say screw you to the hobbyists when the kit leaves enterprise.
|
# ? Dec 14, 2022 14:01 |
|
SolusLunes posted:I was gonna say, refusing self-service updates just sounds like a way to say screw you to the hobbyists when the kit leaves enterprise. They'll still do that by paywalling the files you need
|
# ? Dec 14, 2022 15:01 |
Thanks Ants posted:They'll still do that by paywalling the files you need
|
|
# ? Dec 14, 2022 15:31 |
kzersatz posted:I work with something like 12 pure boxes and one flashblade, I swear if I could set fire to the flashblade and not get sent straight to the poor house, I'd consider it. Ive had a number of friends go to work for pure. All the ones that went flashblade left the company already to a competitor. All the ones in the other product groups are still there.
|
|
# ? Dec 16, 2022 16:20 |
|
Thanks Ants posted:They'll still do that by paywalling the files you need Yeah, but those sometimes make their way out of the paywall. This is just a more robust version of a loving hobbyists solution.
|
# ? Dec 16, 2022 16:36 |
|
BlankSystemDaemon posted:Only company I know of that doesn't do that is Supermicro. Netapp seems to be hit or miss on this- but I also only use a handful of their disk shelves, nothing else.
|
# ? Dec 16, 2022 16:37 |
|
Thanks Ants posted:They'll still do that by paywalling the files you need you can get the files with the work account t
|
# ? Dec 16, 2022 20:18 |
|
Langolas posted:Ive had a number of friends go to work for pure. All the ones that went flashblade left the company already to a competitor. All the ones in the other product groups are still there. They jump to VAST? Honestly, I wouldn't wish Flashblade on my enemies. It's trash, in one calendar year of use, we've easily replaced 3/4 of the blades from failures. Their reasoning? Excessive overwrites. We're using it as a backup target with object lock, if you can't handle the waves stay out of the backup space.
|
# ? Dec 17, 2022 21:49 |
|
Reposting from the storage thread: I am currently messing about with AWS' FSx ONTAP file system and will be building an iSCSI filer to compare/contrast price and performance vs EBS for my specific workload. That said, I'm considering three scenarios for a MSSQL Server machine: 1. one LUN on ONTAP serving one iSCSI disk serving one volume on Windows with a set of databsae files on it (mdf, ndf and log files) 2. one LUN on ONTAP serving one iSCSI disk serving multiple volumes on windows 3. multiple LUNs on ONTAP, each serving one iSCSI disk with a single windows volume on it One would think it wouldn't make a difference because the ONTAP filer is still serving the same about of thoughput, the same amount of storage and the same amount of IOPS. Am I missing something?
|
# ? Jan 26, 2023 23:53 |
|
Haven't really worked with any of that, but from a general storage perspective, here's some of my thoughts. I suspect AWS/NetApp already has specific guidance on this. One of those things that's good to follow, as if there are issues down the road they may just tell you to switch to the correct way if it really gets down to it in troubleshooting. If you ask most DBAs, they'll want things separated out as much as possible, down to the individual hard drives, but lol. What are you using to make the iSCSI connection, Windows iSCSI? I'm not super up to speed on best practices there, but I would check to see limits on the adapter side. More adapters may alleviate some bottlenecks. In my experience, it also depends on performance needs. Covering for the highest performance may add additional complexity that unnecessarily complicates future operations if it's not needed.
|
# ? Jan 27, 2023 00:08 |
|
My company buys capacity from this 3rd party vendor and we're having issue where supposedly two equally specced perform dissimilar, other being almost one fourth slower. I'm going to guess it's due to disk perf (it's database server with Xeon 56c Plats and 128 GB of RAM on both) so what would be quick benchmark to run? Bonnie++ springs to mind but it takes hours to run, is there anything simple and fast that would provide disk metrics we could easily compare?
|
# ? Jan 27, 2023 09:58 |
|
Kivi posted:My company buys capacity from this 3rd party vendor and we're having issue where supposedly two equally specced perform dissimilar, other being almost one fourth slower. I'm going to guess it's due to disk perf (it's database server with Xeon 56c Plats and 128 GB of RAM on both) so what would be quick benchmark to run? Bonnie++ springs to mind but it takes hours to run, is there anything simple and fast that would provide disk metrics we could easily compare? bonnie++ shouldn't take hours to run, or can be config'ed not to. iometer is the standard storage perf characterization tool. You can make it test something very close to your average workload (provided you've characterized that beforehand).
|
# ? Jan 27, 2023 13:09 |
|
Have a look at FIO, that gave me pretty consistent results when I had to do a bunch of storage benchmarks. https://fio.readthedocs.io/en/latest/fio_doc.html The challenge with benchmarking modern storage is that it is really easy to bottleneck on something else since all-flash enterprise storage arrays have become insanely fast. Different tools and test methods give wildly different results which is frustrating. You usually need a combination of large IOs, deep queues or many threads to get anywhere near the max performance of modern storage and there's a ton of badly written software out there that wants to do everything synchronously on a single thread with whatever is the OS defaults for file io.
|
# ? Jan 27, 2023 13:34 |
|
Kivi posted:My company buys capacity from this 3rd party vendor and we're having issue where supposedly two equally specced perform dissimilar, other being almost one fourth slower. I'm going to guess it's due to disk perf (it's database server with Xeon 56c Plats and 128 GB of RAM on both) so what would be quick benchmark to run? Bonnie++ springs to mind but it takes hours to run, is there anything simple and fast that would provide disk metrics we could easily compare? FIO is fine, iometer is fine too, your issue is going to be setting up a proper test to get accurate IO from the test. This means either you already have a profile for the DB (80% read / 20% writes / 8k blocks / etc) or you need to identify the profile you want to use. With that in mind, you also should use a disk twice the size of ram in the system, so if you've got 128 gigs of ram, you should allocate out a 256 gig disk to run the test on. Also, try and keep things the same, if you end up writing the test file on an existing filesystem, make sure it's the same filesystem as the one your db is running on. If it's oracle (and using it's own custom format disk) maybe test on xfs. Anyway, there's lots of things you should be accounting for to run a good test, ram, cpu, and disk not-withstanding. Hope this helps
|
# ? Jan 27, 2023 18:55 |
|
Thanks for the help and suggestions! It's PgSQL on XFS, with separate wal disk and tuning my former co-worker suggested (agcount 56 to match the CPU count) I'll look at bonnie++ again, it just took forever on my testing server but that has "tier-2" disks (tiers go to 0-3, 0 being fastest) so I just assumed it was hours long benchmark. FIO looks like a good tool too.
|
# ? Jan 28, 2023 11:06 |
|
Anyone got any recommendations/horror stories on low-end SAN systems (~100TB/hybrid/rackmount) for data storage? Was looking at something like XiSystems/Truenas X-series kind of thing. A step above homebrew kind of thing.
|
# ? Feb 22, 2023 22:31 |
|
I've never had a supported iX box in my datacenter, but freenas has been reliable for me so it's probably a reasonable choice. The pricing doesn't look too bad over the cost of bare hardware to get support with it. My current benchmark price for enterprise storage is Qumulo, and truenas looks to be 1/3 or less per TB than that.
|
# ? Feb 23, 2023 17:52 |
|
TrueNAS is hard to beat.
|
# ? Feb 24, 2023 01:22 |
|
I got some quotes to replace some old IBM Unified filers. Mixed use - some PACS cache, home directories, random apps, database backups, etc. One of the quotes was for a NetApp C250 with about 200TB. (And another for replication) I can find no reason to get anything else, the price/performance ratio on this thing seems too good to be true. Edit - The other arrays were a NetApp FAS 8300, and EMC Unity 380. Both with mixed disk, ie. not all flash. Edit 2 - I was leaning EMC Unity until this NetApp came along. Kaddish fucked around with this message at 16:45 on Apr 18, 2023 |
# ? Apr 18, 2023 16:39 |
|
Have you gotten a quote on some qumulo archive tier storage? Even that'll do 1GB/sec per node. I'm on my second cluster and the product and support are stellar.
|
# ? Apr 18, 2023 16:51 |
|
|
# ? May 11, 2024 15:25 |
|
Qwijib0 posted:Have you gotten a quote on some qumulo archive tier storage? Even that'll do 1GB/sec per node. I'm on my second cluster and the product and support are stellar. Never even heard of it but I'll check it out!
|
# ? Apr 18, 2023 16:56 |