|
Aquila posted:That was my experience with Hitachi as well.
|
# ? Mar 21, 2014 22:15 |
|
|
# ? May 27, 2024 22:26 |
|
Langolas posted:This is correct. It runs off the management service on the storage processors themselves. Talked to emc/Cdw for more than a hour today in a conference call. They are trying to say that the unisphere is the os of the processing units and is a required charge. They gave a lot of excuses why its a different line item. It seems to be just part of the unit. I don't really care. If they didn't keep trying to sell installation that costs more than 50% of the cost of the solution I wouldn't assume they are such shady assholes. Why are storage salesmen the car dealer salesman of the IT industry? At least getting initial quotes is easy. I got competing quotes from dell, netapp, and emc. Who else should I be talking to?
|
# ? Mar 21, 2014 22:26 |
|
Sickening posted:Talked to emc/Cdw for more than a hour today in a conference call. They are trying to say that the unisphere is the os of the processing units and is a required charge. They gave a lot of excuses why its a different line item. It seems to be just part of the unit. I don't really care. Nimble has some pretty amazing stuff.
|
# ? Mar 21, 2014 22:40 |
|
Sickening posted:At least getting initial quotes is easy. I got competing quotes from dell, netapp, and emc. Who else should I be talking to? Perhaps IBM, as Misogynist said above. He put me in touch with their guys for a project and they were great, it just didn't work out because our needs were too low end. Well, combined with the fact that my boss at the time literally didn't believe in SANs and would rather store critical production data on crappy 1U white-box servers with 4 consumer SATA drives in them. Which were EOL and the only replacement "source" was eBay. But if you're talking seriously with EMC, IBM is not our of your price league
|
# ? Mar 21, 2014 22:46 |
|
Docjowles posted:Perhaps IBM, as Misogynist said above. He put me in touch with their guys for a project and they were great, it just didn't work out because our needs were too low end. Well, combined with the fact that my boss at the time literally didn't believe in SANs and would rather store critical production data on crappy 1U white-box servers with 4 consumer SATA drives in them. Which were EOL and the only replacement "source" was eBay. Our budget is around 35k. I have spent to long running already purchased storage systems and I have only been through one other order before (which I had no influence over at all.) Its my first time on the buying side with actual power. Its amazing how stupid you feel when its your first go around.
|
# ? Mar 21, 2014 22:55 |
|
Sickening posted:Our budget is around 35k. I have spent to long running already purchased storage systems and I have only been through one other order before (which I had no influence over at all.) Its my first time on the buying side with actual power. Its amazing how stupid you feel when its your first go around. For that price I would give a good hard look at Equallogic, unless you absolutely need file services.
|
# ? Mar 21, 2014 23:27 |
|
Sickening posted:At least getting initial quotes is easy. I got competing quotes from dell, netapp, and emc. Who else should I be talking to? Block only? Check out an IBM v3700. If you need file then v7000 Unified.
|
# ? Mar 21, 2014 23:32 |
|
Sickening posted:Our budget is around 35k. I have spent to long running already purchased storage systems and I have only been through one other order before (which I had no influence over at all.) Its my first time on the buying side with actual power. Its amazing how stupid you feel when its your first go around. Are we talking iSCSI or FC here? Do you need any specific features, or is this just a place to dump a ton of data? The Promise VessRAID might be a good place to start; VR2600FIDAME would be both iSCSI (1Gb) and FC (8Gb) in the same box with redundant controllers and 48TB raw HDD capacity for ~15k. http://www.promise.com/media_bank/Download%20Bank/Datasheet/Vess%20R2000%20DS%20v1.6_20130924_en.pdf
|
# ? Mar 21, 2014 23:52 |
|
I once again have to pimp oracle. We love our oracle storage, and the price was pretty good on it as well.
|
# ? Mar 22, 2014 00:44 |
|
adorai posted:I once again have to pimp oracle. We love our oracle storage, and the price was pretty good on it as well. Or this, Oracles stuff is some pretty tight poo poo. You just have to deal with oracle which, well is up to you.
|
# ? Mar 22, 2014 00:53 |
|
Dilbert As gently caress posted:Or this, Oracles stuff is some pretty tight poo poo. You just have to deal with oracle which, well is up to you.
|
# ? Mar 22, 2014 01:00 |
|
KillHour posted:Are we talking iSCSI or FC here? Do you need any specific features, or is this just a place to dump a ton of data? FC . Its what I know and should meet our needs for as long as we own the next system. Tiering has always been something I have liked to play with and make sense for what we do. We are a small shop that appears to be growing at a pretty decent pace. We only have a NAS right now and everything else is local storage. We are talking 3 vmware hosts with about 15 guests. Most of those guests are SQL databases that run some pretty heavy production apps.
|
# ? Mar 22, 2014 01:06 |
|
Dilbert As gently caress posted:Nimble has some pretty amazing stuff. This. If you want some raw block iscsi only storage, this is a great bet. Replication is amazing as well. My only complaint is it so simple. Bores me. Edit. FC rules this out. Why do you prefer FC? A lot invested on the switching already? Moey fucked around with this message at 02:05 on Mar 22, 2014 |
# ? Mar 22, 2014 01:25 |
|
Sickening posted:FC . Its what I know and should meet our needs for as long as we own the next system. Tiering has always been something I have liked to play with and make sense for what we do. So you're looking at a VNXe 3300 w/ FAST cache, and FC connections? My next phone call would be Dell the MD32/36's can do a similar fast cache and are price competitive, HP storage is poo poo. Also FC is bleh, I really dislike working with FC. Moey posted:This. Only thing I don't like about nimble is that it's ISCSI only and view (correct me if I am wrong) likes only to support recomposes and provisions off NFS VAAI. Yes I realize the backend arch, but the offloading is nice when you're talking 3k desktops. Dilbert As FUCK fucked around with this message at 01:54 on Mar 22, 2014 |
# ? Mar 22, 2014 01:36 |
|
The big problem with Nimble is that it's likely that they won't exist in any meaningful form in five years time. The storage hardware market is brutal to startups and Nimble has a) not made any money and b) invested heavily in a technology with an expiration date in the not too distant future. Not turning a profit isn't really a big deal for a startup in their current position, but the road to profitability seems pretty tricky for them since their major leverage is low cost and at some point they are going to have to stop giving their hardware away to turn a profit. Their financials look a lot like Violin, which tanked pretty shortly after it's IPO. But their stock has been mostly stable since the IPO (not a very long time, but longer than Violin) so they may be on better footing there. The bigger problem is that hybrid arrays are a bridge technology that are going to be killed off by all flash arrays and software defined storage. Their benefits are predicated on SSD being really expensive and have terribly capacity and both of those things are not going to remain true for long enough for them to establish a foothold that will keep them alive long enough to develop a new product. Of course if you just want cheap fast storage to last the next 3 years or so, and don't mind that you might very well have to replace it with something completely different from another vendor at your next refresh cycle then it's a pretty hard proposition to beat. But the odds of any storage startup making it are really really slim. 6 companies own more than 80% of the storage marketshare and the youngest of those is 22 years old. Only two of the six started life as storage systems manufacturers. It's a really capital intensive business and it gets significantly more challenging the larger you get. Add to that the fact that VMWare and Microsoft are trying to put dedicate storage vendors out of business and things like Atlantis USX and other SDS solutions are driving commoditization and it gets even more difficult to see how a traditional storage startup survives. That's not to say that Nimble, Tintri, Nutanix, and some of these other startups don't have really cool products, I just question their longevity.
|
# ? Mar 22, 2014 05:55 |
|
NippleFloss posted:b) invested heavily in a technology with an expiration date in the not too distant future.
|
# ? Mar 22, 2014 06:15 |
|
I think SomethingAwful is the only place people even talk about Nimble anymore. They should pay you guys.
|
# ? Mar 22, 2014 06:16 |
|
three posted:I think SomethingAwful is the only place people even talk about Nimble anymore. Just wait till cisco buys them anyways, then it's poo poo just like 3Par. I'm just glad ZFS is being adopted. I don't like nimbles ISCSI only but it works well. NippleFloss posted:The big problem with Nimble is that it's likely that they won't exist in any meaningful form in five years time. If cisco wasn't making on them so heavily I'd agree but I fully expect a cisco buy out in the next 4 years. Dilbert As FUCK fucked around with this message at 06:21 on Mar 22, 2014 |
# ? Mar 22, 2014 06:18 |
|
Dilbert As gently caress posted:Just wait till cisco buys them anyways, then it's poo poo just like 3Par. While Invicta sucks, I doubt they're going to give up on it that quickly.
|
# ? Mar 22, 2014 06:20 |
|
three posted:I think SomethingAwful is the only place people even talk about Nimble anymore.
|
# ? Mar 22, 2014 06:21 |
|
three posted:While Invicta sucks, I doubt they're going to give up on it that quickly. I'd dig up the article but Cisco is heavily invested in nimble
|
# ? Mar 22, 2014 06:22 |
|
Dilbert As gently caress posted:I'd dig up the article but Cisco is heavily invested in nimble #ThingsFrom2012
|
# ? Mar 22, 2014 06:25 |
|
adorai posted:I have serious doubts that all flash arrays are coming in the not too distant future. I think flash accelerated spinning disk is going to be dominant array type for many years. They're already here. Every major storage company has an AFA in it's portfolio now and most are either looking to either build or acquire purpose built AFAs. Pure is probably a larger company than Nimble at this point, though it's hard to tell since they're still private. They have a larger workforce and a very mature offering. Revenue for the AFA market is going to crest $1 billion by 2015 at the current rate. SSD sizes have already reached parity with 10k and 15k HDD sizes, with 1.2TB SSDs available from some vendors. Cost is the only limiting factor and it will continue to come down. HDD isn't going to disappear entirely, nor will hybrid arrays, but company's that rely on mainly on being fast and cheap will have a hard time competing when AFAs mean that almost everything is faster than most customers will ever need. Then they have to win on price, and a small company like Nimble can't undercut bigger companies with more resources indefinitely. SDS will be the final nail in the coffin, with hybrid like functionality being layered on top of commodity hardware or disparate arrays via software, rather than being backed into the hardware. NetApp is worried about Nimble in the short term because we lose deals to them right now, but longer term the company is much more worried about Pure, Atlantis, Nutanix, VSAN, and truly disruptive technologies. Hybrid is pretty easy and basically everyone does it now. That other stuff is where the market will really differentiate. Dilbert As gently caress posted:I'd dig up the article but Cisco is heavily invested in nimble Cisco created enough problems for themselves with the Whiptail purchase and they have been bending over backwards to assure their storage partners that they have no intention of competing with them directly. If they bought a direct competitor to EMC and Netapp and the other traditional storage vendors the clamor would reach the heavens. This also ignores that they still haven't integrated Whiptail technology at all, and if they were going to purchase storage it would very likely be a converged technology like Nutanix that has a natural synergy with their UCS hardware, and not a standalone array. Cisco is the only suitor that makes any sense for Nimble, but I don't think they make that much sense either.
|
# ? Mar 22, 2014 07:11 |
|
Sickening posted:FC . Its what I know and should meet our needs for as long as we own the next system. Tiering has always been something I have liked to play with and make sense for what we do. I'd suggest look at Hitachi Data Systems (HDS) and their HUS range. I'm also storage shopping and no kidding, they are so refreshingly free of all the bullshit I've had to deal with from other vendors that it's been a breath of fresh air so far. You haven't mentioned IOPS but something like a HUS 110 with dual symmetric FC controller and 24x 900GB 10K spindles comes out at around $25K with three years support (and that's an initial quote without trying to work on discount levels) Direct connect over FC for 3 hosts so no FC switching needed and it can do all the funky stuff like tiering and flash if and when you need it.
|
# ? Mar 22, 2014 10:47 |
|
Dilbert As gently caress posted:Just wait till cisco buys them anyways, then it's poo poo just like 3Par. Iunno, the big honking 3Par V800 that one of our sister companies got to run Postgres on managed to convince people that SANs actually might be useful instead of something to be shunned.
|
# ? Mar 23, 2014 17:22 |
|
Sickening posted:Talked to emc/Cdw for more than a hour today in a conference call. They are trying to say that the unisphere is the os of the processing units and is a required charge. They gave a lot of excuses why its a different line item. It seems to be just part of the unit. I don't really care. The reason it is a different line item is very simple. A lot of the vendors these days are striving to show more software sales. So where Unisphere used to be bundles into the cost of the array hardware it is now split out to show a greater percentage of software sales. When it comes to services that's where the likes of CDW are desperate to sell to you - it's where they make their margin - they get very little from the array sale itself. Sell a chunk of services and they may get 5x the margin than from the array sale even though it's half the cost.
|
# ? Mar 23, 2014 19:02 |
|
Bitch Stewie posted:Direct connect over FC for 3 hosts so no FC switching needed and it can do all the funky stuff like tiering and flash if and when you need it. Yeah, which is exactly what I ended up doing with my HUS110 after months of headaches, troubleshooting, and thousands of dollars in switches couldn't resolve our mysterious latency spikes on the VMware datastores when hooked up via iSCSI...
|
# ? Mar 26, 2014 14:54 |
|
Protip: If you svmotion to and from datastores on a SIS NetApp volume, check your SIS database. We found a 750GB database on a datastore with 600GB of VMs.
|
# ? Mar 28, 2014 23:36 |
|
If we're sharing awesome NetApp issues, we got hit by the memory leak descrbied in this post recently. Seeing random spikes in the load on one filer and our app's perf completely goes to poo poo til it recovers. Sure enough, there's a corresponding "WAFL is running low on memory" message in the log every time our monitoring system blows up with perf alerts.
|
# ? Mar 28, 2014 23:44 |
|
Do people still buy NetApp?
|
# ? Mar 29, 2014 20:43 |
|
three posted:Do people still buy NetApp? Yes.
|
# ? Mar 30, 2014 04:21 |
|
1000101 posted:Yes. Do these people feel shame?
|
# ? Mar 30, 2014 04:22 |
|
three posted:Do these people feel shame? Not at all. It's still a solid and pretty easy to use product. I would buy it if I could.
|
# ? Mar 30, 2014 04:26 |
|
1000101 posted:Not at all. It's still a solid and pretty easy to use product. I would buy it if I could. you silly person
|
# ? Mar 30, 2014 05:12 |
|
Obviously the correct answer is to buy <array from established vendor that is easy to manage and cheap and problem free>. I have compiled a list of all such arrays:
|
# ? Mar 30, 2014 05:57 |
|
^^ haha NetApp has been good for us. It's not cheap or without bugs but what storage is? We haven't experienced anything like the horror stories I hear about EMC.
|
# ? Mar 30, 2014 07:53 |
|
so on Tuesday we had an IOXM (IO expansion module) die on controller A of our HA pair. Today we had the mainboard of controller B die. This was after both controllers had over 600 days of uptime. I do not believe that these module actually died and that it is instead one of the shelves that is causing the problem, but I have to play the game with support I suppose.
|
# ? Mar 31, 2014 02:24 |
|
three posted:Do people still buy NetApp? Any storage solution is only as good as how you design it. Netapp can do amazing things with NFS, and is always a good second option to EMC.
|
# ? Mar 31, 2014 05:27 |
|
Dilbert As gently caress posted:Any storage solution is only as good as how you design it. This is a pretty stupid glittering generality. NetApp has been falling behind for quite some time, both technologically (e.g. still no equivalent product to XtremIO/Pure, and from people exiting the company (e.g. Vaughn Stewart). They've plateau'd and will be in some serious trouble if they don't get their poo poo together.
|
# ? Mar 31, 2014 13:41 |
|
|
# ? May 27, 2024 22:26 |
|
Dilbert As gently caress posted:Any storage solution is only as good as how you design it. Netapp can do amazing things with NFS, and is always a good second option to EMC. I've only started working with NetApp recently and unfortunately my experience has been soured by having to support a system running pre-8.2 Clustered Data ONTAP (If memory serves it was deployed running 8.0.2 by a third-party at the customer's behest). We are halfway towards upgrading to 8.2, currently at the mid-point having upgraded to 8.1.3 and upgrading SnapDrive on the hosts. Our progress has been hampered somewhat by our completely inept technical architects who when deploying 14 new ESXi hosts managed to configure two of them with the same iSCSI initiator IP. Our vSphere environment is 90% NFS and 10% iSCSI so it went unnoticed for about a week until DRS migrated a VM running on iSCSI datastores to one of the misconfigured hosts. That was a fun day All that BS aside, NetApp seems OK performance-wise, at least for our workloads.
|
# ? Mar 31, 2014 15:46 |