|
NetApp E-Series or IBM v3700
|
#
?
Jun 2, 2016 07:57
|
|
|
|
| # ? Apr 19, 2024 07:54 |
|
|
Thanks Ants posted:NetApp E-Series or IBM v3700 goobernoodles fucked around with this message at 16:16 on Jun 2, 2016 |
|
#
?
Jun 2, 2016 16:13
|
|
|
All of our new Sans for the new colo and dr and accompanying switches have shipped, as soon as we have network up at the end of next week in the colo it's go time. 
|
|
#
?
Jun 2, 2016 16:27
|
|
|
Thanks Ants posted:So NetApp seem quite proud of ONTAP 9. Does anyone have a quick rundown of what it's all about? They seem to be focusing on all-flash quite a lot - is this because it brings little to the table unless you're running all-flash, or are they just emphasizing it because all-flash is A Thing right now? There are some nice things in ONTAP 9. Data compaction may help mitigate the fixed 4K block size and amplify the benefit of inline compression, which currently isn't competitive with variable block vendors like Nimble and Tegile. Scale out file systems will be useful for customers who want an Isilon alternative, though I don't think it will ever get as cheap as Isilon can. Supposedly there are some nice performance improvements as well. ONTAP Select should be interesting since software defined storage is picking up momentum and bespoke hardware will get tougher and tougher to sell. The push to market all flash is because the economics of flash are such that all flash will make more sense than hybrid for a lot of customers. A lot of people have bought the last spinning disk array they will ever buy. Getting to the front of the all flash market is vital for any storage company if they want to succeed. NetApp will support 15TB SSD drives on ONTAP 9, which are notably larger than even the largest SATA drives. The capacities are there to put everything on flash and the cost is getting there.
|
|
#
?
Jun 2, 2016 18:40
|
|
|
PCjr sidecar posted:Optics, but the official Chelsio optics are cheap. We've used twinax elsewhere; it's OK, but it's occasionally a crapshoot going switch->host or between different vendors' switches. Also, gently caress Emulex.
|
|
#
?
Jun 3, 2016 08:18
|
|
|
Vulture Culture posted:Oh hey, just noticed they have copper too. Have you used the twinax from FS.com? I'm curious about this as well. Have $1400 worth of DAC from there in a cart and have been waiting to pull the trigger
|
|
#
?
Jun 4, 2016 15:01
|
|
|
I was thinking of joining SNIA to fluff up my resume a bit, but 300 clams is a lot to ask. Anyone know if (individual) membership is worthwhile overall? I'm going to peruse their site for more information, I was just curious if anyone had personal good/bad experiences.
|
|
#
?
Jun 6, 2016 16:36
|
|
|
Seventh Arrow posted:I was thinking of joining SNIA to fluff up my resume a bit, but 300 clams is a lot to ask. Anyone know if (individual) membership is worthwhile overall? I'm going to peruse their site for more information, I was just curious if anyone had personal good/bad experiences. Not worth it. I've spent a lot of time in the storage industry and never met anyone who cared at all about SNIA.
|
|
#
?
Jun 6, 2016 19:24
|
|
|
Good to know, thanks!
|
|
#
?
Jun 6, 2016 19:30
|
|
|
No experience with their DACs. Our networking team stocks OEM DACs but fs.com optics; I'm not sure if that's just because DACs are a low-volume product for them or a result of a specific experience with a fs.com DAC. I've had fs.com cancel an order for a QSFP active optic cable in a weird configuration because they couldn't validate it in their test lab and hadn't shipped any in that configuration, so they're thinking about compatibility.
|
|
#
?
Jun 6, 2016 20:38
|
|
|
Anyone have experience with Veeam backups directly from VNX storage snapshots? Its a new feature in Veeam 9 and we are thinking about adding snapshot licensing to the new VNX we're buying.
|
|
#
?
Jun 7, 2016 15:32
|
|
|
ZetsurinPower posted:Anyone have experience with Veeam backups directly from VNX storage snapshots? Its a new feature in Veeam 9 and we are thinking about adding snapshot licensing to the new VNX we're buying. I am interested in this as well, but via Nimble snaps. Looking to switch to Veeam as well. PHD Virtual got bought by Unitrends, and I have been an unhappy camper ever since.
|
|
#
?
Jun 7, 2016 15:37
|
|
|
Veeam is pretty great, but it would be truly amazing to eliminate stun periods. This would let us run backups during production without any interruptions. Very cool stuff.
|
|
#
?
Jun 7, 2016 16:04
|
|
|
Moey posted:I am interested in this as well, but via Nimble snaps. Looking to switch to Veeam as well. PHD Virtual got bought by Unitrends, and I have been an unhappy camper ever since. I made the jump from PHD Virtual to Veeam about 6 months ago and could not be happier. The only things I have run into, and I admit, they are big ones, are overshadowed by how good the rest of the product is. Make sure you account for how Veeam handles Grandfather, father, son (GFS) retention, because if you want to keep something like 14 dailies, 4 weeklies, 12 monthlies, and X yearlies, you are going to need a ton of storage space or something that can do dedupe. https://helpcenter.veeam.com/backup/vsphere/backup_copy_gfs.html https://helpcenter.veeam.com/backup/vsphere/backup_copy_gfs_cycles.html Oh, and if you are looking at implementing "Direct SAN" backups, make sure your SAN supports read-only iSCSI access. Otherwise you can overwrite production data way too easily for my comfort. I chose not to use it.
|
|
#
?
Jun 7, 2016 19:32
|
|
|
Our retention here is a joke, 90 days of nightly backups. Never anything more. I used Veeam in the past, and inherited PHD then Unitrends at my current place. Running Nimble for production (and cross site replication) then some Dell MD3200i filled with 4tb drives for on-site backups.
|
|
#
?
Jun 7, 2016 20:40
|
|
|
If the Nimble integration is functionally identical to the NetApp integration it will be very good. If you're fully virtual and not massive VEEAM is the best thing out there.
|
|
#
?
Jun 7, 2016 21:18
|
|
|
Does veeam give me anything that a second SAN to replicate to doesn't?
|
|
#
?
Jun 8, 2016 01:13
|
|
|
adorai posted:Does veeam give me anything that a second SAN to replicate to doesn't?
|
|
#
?
Jun 8, 2016 01:17
|
|
|
adorai posted:Does veeam give me anything that a second SAN to replicate to doesn't? My absolute last resort for replicating data is SAN replication.
|
|
#
?
Jun 8, 2016 02:10
|
|
|
Internet Explorer posted:My absolute last resort for replicating data is SAN replication.
|
|
#
?
Jun 8, 2016 02:27
|
|
|
adorai posted:Does veeam give me anything that a second SAN to replicate to doesn't? Vulture Culture posted:Per-file/item restores are a big one in certain use cases, along with per-VM differentials (presuming you don't keep one VM per LUN right now) Also the ability to fail over a single VM without having to break the mirror and bring the whole mirrored datastore online. And VEEAM replication will take car of adding the VM to inventory so that it's ready to be turned up immediately. It can also handle doing some basic re-IP work and association with different port groups if required, without needing SRM to do that.
|
|
#
?
Jun 8, 2016 02:33
|
|
|
adorai posted:Does veeam give me anything that a second SAN to replicate to doesn't? A really nice way to get your data out? Restore it? Per object AD restore, record sql, item exchange without needing agents? Automatically testing your replicated VMs? It is very good.
|
|
#
?
Jun 8, 2016 05:46
|
|
|
What's peoples goto these days for big, cheap, reliable archival NFS storage. Async site to site replication would be great too, just really don't have the time to CJ my own solution for archival storage. Our primary is Tegile, looking for something with a per GB cost somewhere between that and a big fat QNAP. 
|
|
#
?
Jun 9, 2016 00:18
|
|
|
Data Domain would be my go-to for that but I'm interested in the options as well. A FAS 2520 filled with capacity disks isn't hugely expensive either.
|
|
#
?
Jun 9, 2016 00:33
|
|
|
Maneki Neko posted:What's peoples goto these days for big, cheap, reliable archival NFS storage. Async site to site replication would be great too, just really don't have the time to CJ my own solution for archival storage.
|
|
#
?
Jun 9, 2016 01:16
|
|
|
Funny, our company too is looking to switch to something less expensive than our current 300TB Isilon cluster after receiving a renewal quote that our budget couldn't match. Need something in between that and just using a rather large VM or two to store all of our NFS data on Nimble.
|
|
#
?
Jun 9, 2016 02:06
|
|
|
Tegile, Oracle ZFS appliances, or NetApp are your cheapest options for enterprise quality NFS storage.
|
|
#
?
Jun 9, 2016 03:11
|
|
|
Maneki Neko posted:What's peoples goto these days for big, cheap, reliable archival NFS storage. Async site to site replication would be great too, just really don't have the time to CJ my own solution for archival storage.
|
|
#
?
Jun 9, 2016 09:20
|
|
|
NippleFloss posted:Tegile, Oracle ZFS appliances, or NetApp are your cheapest options for enterprise quality NFS storage. We already run Tegile for primary storage and I don't think they can get pricing down where I'd like, so maybe it is time to hit up our local Netapp guy and see how hungry he is. Oracle is a 4 letter word here after they spent the last year trying to gently caress us super hard on licensing on the software side. evil_bunnY posted:Netapp just made us an offer we literally could not refuse. I can PM you the pricing if you'd like. I don't know what tegile costs (they have no VARs here). Sure, that would be great! Maneki Neko fucked around with this message at 16:15 on Jun 9, 2016 |
|
#
?
Jun 9, 2016 16:13
|
|
|
I don't work in my big company's data center but I wanted to guess how much 625T of slow(ish) enterprise storage would cost. Something that could input/output maybe 2-4G across 4 simultaneous hosts max in practice. For large files. I too embarrassed to show my lovely math but I'm guessing $150-$200k for some turn key product?
|
|
#
?
Jun 9, 2016 19:34
|
|
|
Shaocaholica posted:I don't work in my big company's data center but I wanted to guess how much 625T of slow(ish) enterprise storage would cost. Something that could input/output maybe 2-4G across 4 simultaneous hosts max in practice. For large files. I too embarrassed to show my lovely math but I'm guessing $150-$200k for some turn key product? You could get to that price point/capacity with some cheap secondary block storage on NLSAS pretty easily.
|
|
#
?
Jun 9, 2016 19:43
|
|
|
evil_bunnY posted:Netapp just made us an offer we literally could not refuse. I can PM you the pricing if you'd like. I don't know what tegile costs (they have no VARs here). I would be interested to know this as well!
|
|
#
?
Jun 9, 2016 19:54
|
|
|
Shaocaholica posted:I don't work in my big company's data center but I wanted to guess how much 625T of slow(ish) enterprise storage would cost. Something that could input/output maybe 2-4G across 4 simultaneous hosts max in practice. For large files. I too embarrassed to show my lovely math but I'm guessing $150-$200k for some turn key product? 4 gigabit or gigabyte per second? Cause 4GB/s isn't exactly slow.
|
|
#
?
Jun 9, 2016 20:34
|
|
|
NippleFloss posted:4 gigabit or gigabyte per second? Cause 4GB/s isn't exactly slow. Gigabit. Like 4 simultaneous 1G connections to 4 clients.
|
|
#
?
Jun 9, 2016 23:11
|
|
|
NippleFloss posted:4 gigabit or gigabyte per second? Cause 4GB/s isn't exactly slow.
|
|
#
?
Jun 9, 2016 23:13
|
|
|
Shaocaholica posted:Gigabit. Like 4 simultaneous 1G connections to 4 clients. What protocol will the hosts use to access the storage? Vulture Culture posted:Depends on workload. Even assuming big slow 4 TB NLSAS disks, that's at least 175 spindles. If you're streaming large writes and striping wide, you should conservatively be able to hit 15 GB+ with a good enough write-back cache. Not that anyone, you know, would. At that data rate the controller becomes a bottleneck. The highest end NetApp controller can push about 1800MB/s through a single controller with a 64KB all sequential write workload. A Nimble C700 is going to be around that number as well. E-Series or Isilon or something along those lines would work better, but I'd hesitate to call any of those options slow. They're just fast in a different way than an all flash array is fast. Multiple GB/s of throughput from just a few clients is still pretty fast in the traditional storage space. Not so much for big data I guess, but those guys aren't buying traditional storage.
|
|
#
?
Jun 10, 2016 01:00
|
|
|
NippleFloss posted:At that data rate the controller becomes a bottleneck. The highest end NetApp controller can push about 1800MB/s through a single controller with a 64KB all sequential write workload. A Nimble C700 is going to be around that number as well. E-Series or Isilon or something along those lines would work better, but I'd hesitate to call any of those options slow. They're just fast in a different way than an all flash array is fast. Multiple GB/s of throughput from just a few clients is still pretty fast in the traditional storage space. Not so much for big data I guess, but those guys aren't buying traditional storage. Of course, if it was really a single sequential workload like my spitballing, you'd need 100GbE+ to hit those numbers. So yeah, if you're not basically generating that data locally, it's still pretty expensive because you're talking about a number of controllers around a clustered filesystem. 
Vulture Culture fucked around with this message at 01:43 on Jun 10, 2016 |
|
#
?
Jun 10, 2016 01:41
|
|
|
Welp, pulled the trigger on a 3x Cisco UCS C220 M4 servers w/ 2630v3's, 1.2Tb of RAM and a Tegile T3530 along with a 2x servers and a T3100 array for our 2nd office. Cha-ching. e: Well, that and a HP 5406R for the 2nd office and 10Gb SFP+ modules for both offices. Should be quite the upgrade from our 8 year old x3650's and DS3300. goobernoodles fucked around with this message at 04:28 on Jun 11, 2016 |
|
#
?
Jun 11, 2016 04:07
|
|
|
Why the Cisco C series vs rando rack mount boxes from HP/Dell, out of curiosity? Did they come in with a super competitive price? UCS is cool but the extreme flexibility seems kinda wasted, and like it's just adding complexity/points of failure on a three node deployment. Not trying to be a nitpicky rear end in a top hat; genuinely interested
|
|
#
?
Jun 12, 2016 20:12
|
|
|
|
| # ? Apr 19, 2024 07:54 |
|
|
Docjowles posted:Why the Cisco C series vs rando rack mount boxes from HP/Dell, out of curiosity? Did they come in with a super competitive price? UCS is cool but the extreme flexibility seems kinda wasted, and like it's just adding complexity/points of failure on a three node deployment. They're likely using the C series as standard rack mount boxes, and not part of a UCS managed deployment. C series can happily run as standard dumb servers without UCS Manager or fabric interconnects.
|
|
#
?
Jun 12, 2016 20:23
|
|