Enterprise Grade Storage Recommendations

[NobleX13]

We are a VMware shop with about 8,000 employees and I am looking for recommendations as to what the "best" storage solution for our use case. We have an IBM XIV Gen 4 SAN right now and I am not exactly a fan.

We have no plans to move to a Nutanix-style hyper-converged environment just yet. We just spent some money on new HP blades a year or so ago. As such, I am looking for a stand-alone solution that I can simply connect to our VMware environment and start creating new datastores on.

My previous employer used a combination of huge Hitachi arrays and Netapp.

We have ~150TB of disk right now, but I would want the new solution to have at least 500TB to start.

 

[kdh]

Budget? I'm pretty much an EMC Fan boy..

You can easily get to 500Ts on a VNX 7600 with Flash Cache, and storage tiring using SSD, SAS 10k drives and 4TB sata drives. But that's a rough stab at it. What kinda data is running in the VMs? VDI, any kind of high io, db type stuff? You can go with xtremio, you'll most likely get 2:1 dedupe and compression so you could go with 6 node brick. Both of those do FC.. if FC is not going to work, Isilon with a few accelerator nodes using NFS over 10gig.

 

[ND40oz]

Been happy with the NetApp AFF, once you go all flash, you're not going to want to go back to spinning platters. Management hasn't changed much if you've used them before except I believe you have to use cluster mode and not 7 mode.

 

[Blue Fox]

If you don't need primarily flash, 3PAR 7400? Was pretty happy with them in our large VM environment at my previous employer. Performance was good and they were quite easy to manage.

 

[NobleX13]

Budget? I'm pretty much an EMC Fan boy..

You can easily get to 500Ts on a VNX 7600 with Flash Cache, and storage tiring using SSD, SAS 10k drives and 4TB sata drives. But that's a rough stab at it. What kinda data is running in the VMs? VDI, any kind of high io, db type stuff? You can go with xtremio, you'll most likely get 2:1 dedupe and compression so you could go with 6 node brick. Both of those do FC.. if FC is not going to work, Isilon with a few accelerator nodes using NFS over 10gig.

I like the sound of that VNX 7600. We are not super IOPS-heavy, although we do have a few hundred VDI desktops. FC would totally work. We have EMC Data Domain and Avamar already, but support has not left the best impression this week. One of the Avamar storage nodes failed and it has taken them a week to restore service.

Been happy with the NetApp AFF, once you go all flash, you're not going to want to go back to spinning platters. Management hasn't changed much if you've used them before except I believe you have to use cluster mode and not 7 mode.

I agree with the all-flash argument. I could re-use our existing SAN to house unstructured user data (file shares) and keep the rest on flash. I don't know what the budget for this stuff is.

If you don't need primarily flash, 3PAR 7400? Was pretty happy with them in our large VM environment at my previous employer. Performance was good and they were quite easy to manage.

That is also a solid option. Our only storage guy is out of the office for six weeks and nobody can tell me why we spent so much money with IBMs rather... unusual product offering.

 

[delita]

Look into Pure Storage...big wigs from NetApp and EMC that left to form a new company based entirely on flash storage and super high data compression. I've sold a couple of their boxes to customers...they have the highest netpromoter score in IT (higher than Apple) and a 30 day money back guarantee.

 

[LazN]

Personally I am a fan of Compellents, easy to use and scale pretty well. And configured right can really perform. But I want to try some of the hyperconverged solutions next. (edit: and with the DELL-EMC merger no one knows yet what will become of Compellent)

 

[kdh]

I like the sound of that VNX 7600. We are not super IOPS-heavy, although we do have a few hundred VDI desktops. FC would totally work. We have EMC Data Domain and Avamar already, but support has not left the best impression this week. One of the Avamar storage nodes failed and it has taken them a week to restore service.

VNX support is a different division then DD or Avamar. Actually DD support is pretty good with the issues I've had. Very smart folks.

 

[kdh]

Look into Pure Storage...big wigs from NetApp and EMC that left to form a new company based entirely on flash storage and super high data compression. I've sold a couple of their boxes to customers...they have the highest netpromoter score in IT (higher than Apple) and a 30 day money back guarantee.

Disagree. I have Pure, and I'm not impressed. My sales guys and TC are shady as all hell. I've caught them lying to me about the capabilities of their product 4 times. Over exaggerating Dedup and Compression, Not being fully transparent on the post compression process, not being fully transparent on when that process is shut off because the array is over loaded, the documentation on their own website, oracle, and VMware saying something completely different then what they say in meetings. Migrating from EMC powerpath to Pure on a Linux host means a Linux host reload but they will tell you something different. No local user authentication other then a single pureuser account? whaaaaat? The boxes I demo'd were basically rebranded Dell boxes running Centos. The next gen hardware was nicer and more modular, but still based on centos. Further more, the sales team will underspecify what you need, so when it gets on your floor and you ram it into the ground, you'll have to upgrade it. Granted swapping engines is easy and should be seamless, but you are taking down half your storage compute power while you do that. Pure will tell you, plenty of people are running away from EMC to go to Pure.. But what Pure doesn't tell you is how many people are running away from Pure to go back to EMC. Oh and the kicker? They will murder the entry level price to get it on your floor.. But when it comes time for upgrades, they really stick it to you. If you compare Pure with XtremIO in terms of technology, hardware, layout, design, support, management, upgradability and TCO? Xtremio walks all over Pure all day everyday. If you look at Pure, take a second look at XtremIO.. Because while the cost to get into XtremIO maybe more expensive, the total cost of owning it will be about the same as Pure in 4 years. How do I know? I'm living it as I type this.

Oh.. and with questions on wether or not Pure is actually profitable as a company is a huge red flag.

Competitors And Cash Bleed Put Pressure On Pure Storage

Dunno about you, but that's not good.. But the rumor mill says while they are a public company, they a looking to be swallowed up by a Cisco for UCS, or Oracle for exadata.

 

[diizzy]

Hitachi (HDS) works well AFAIK, Gandi.net just made a talk at BSDCan about how they moved away from Netapp to FreeBSD depending on what you're looking for...

BSDCan2016: FreeBSD based high density filers

 

[lcpiper]

Been happy with the NetApp AFF, once you go all flash, you're not going to want to go back to spinning platters. Management hasn't changed much if you've used them before except I believe you have to use cluster mode and not 7 mode.

7-Mode still works but they are coming out with a new OnTap solution that combines both capabilities of 7-Mode and C-Dot. I was told it will just be called Data OnTap. Surprised right

I mostly have NetApp systems here running 7-Mode because we just don't need C-Dot yet. Chances are we will move to the new OS from 7-Mode because when that day comes it will be the only tune NetApp is whistling.

That being said I have worked in 3-PAR environments and as long as you are comfortable with iSCSI LUNs it should all go fine. EMC has been getting good reception.

Just make sure you need flash before you push to purchase it. Those that need it really benefit from it, and those that don't have wasted their money and that money could do much more elsewhere. Like all things right.

I'd push NetApp at you cause that's what I am comfortable with, and a contractor never knows when he'll need that next job

 

[Olga-SAN]

what issues did you have with their dedupe? any data loss?

VNX support is a different division then DD or Avamar. Actually DD support is pretty good with the issues I've had. Very smart folks.

 

[Schmeisser]

I typically hear about Hitachi/IBM at the scale you mentioned.
I've got a customer with a mixed environment @ 130 TB, while scale is smaller the approach may interesting.
They're using StarWind's HCA for their 4-node View VDI. But, the remaining infrastructure is not HCA, it is 2 dedicated storage servers + 6-node cluster running rest of the VMs. Storage software running on storage servers is also StarWind. Re-purposed old FC SAN connected to the backend and then presented by StarWind as NFS. One vendor for storage, one vendor for virtualization.

 

[lcpiper]

I like separating NFS exports from File Shares and even iSCSI when I can, at the physical. But money usually dictates otherwise if you don't have the scale to justify it.

 

[Space_Ranger]

What's the estimated price tag on that VNX 7600? We currently have a Thecus N16000V for our video archive (with 3 D16000's hanging off of it. 200TB of total storage so far). Looking for a more efficient way of dealing with our even-increasing storage needs.

 

[CombatChrisNC]

Disagree. I have Pure, and I'm not impressed. My sales guys and TC are shady as all hell. I've caught them lying to me about the capabilities of their product 4 times. Over exaggerating Dedup and Compression, Not being fully transparent on the post compression process, not being fully transparent on when that process is shut off because the array is over loaded, the documentation on their own website, oracle, and VMware saying something completely different then what they say in meetings. Migrating from EMC powerpath to Pure on a Linux host means a Linux host reload but they will tell you something different. No local user authentication other then a single pureuser account? whaaaaat? The boxes I demo'd were basically rebranded Dell boxes running Centos. The next gen hardware was nicer and more modular, but still based on centos. Further more, the sales team will underspecify what you need, so when it gets on your floor and you ram it into the ground, you'll have to upgrade it. Granted swapping engines is easy and should be seamless, but you are taking down half your storage compute power while you do that. Pure will tell you, plenty of people are running away from EMC to go to Pure.. But what Pure doesn't tell you is how many people are running away from Pure to go back to EMC. Oh and the kicker? They will murder the entry level price to get it on your floor.. But when it comes time for upgrades, they really stick it to you. If you compare Pure with XtremIO in terms of technology, hardware, layout, design, support, management, upgradability and TCO? Xtremio walks all over Pure all day everyday. If you look at Pure, take a second look at XtremIO.. Because while the cost to get into XtremIO maybe more expensive, the total cost of owning it will be about the same as Pure in 4 years. How do I know? I'm living it as I type this.

Oh.. and with questions on wether or not Pure is actually profitable as a company is a huge red flag.

Competitors And Cash Bleed Put Pressure On Pure Storage

Dunno about you, but that's not good.. But the rumor mill says while they are a public company, they a looking to be swallowed up by a Cisco for UCS, or Oracle for exadata.

We currently have a VNX5200 and it's closing in on both capacity and throughout. We're thinking about going Pure. What product of theirs are you using exactly? Flashblade looks very impressive to me, compared to the VNX5600 we could get at near the same price.

I'm somewhat new to managing enterprise storage, but an AFA is very appealing.

This would be part of a large data-center upgrade including moving from 7 VMware hosts with dual hex-cores and 192GB of RAM to a fully populated Cisco UCS with 512GB of RAM per blade and the 22 core Xeon V4's.

 

[Shockey]

For those of you considering VNX series might want to look into Unity solutions also. It the next line of VNX series. Of course it has all flash option too.

What's the estimated price tag on that VNX 7600? We currently have a Thecus N16000V for our video archive (with 3 D16000's hanging off of it. 200TB of total storage so far). Looking for a more efficient way of dealing with our even-increasing storage needs.

Isilon product would be worth looking at. Scales with ease. NFS and CIFS access at 10GIG.

 

[Nate7311]

Be careful with depending on dedupe in calculations in SANs. It's typically not inline and can kill performance when the Dedupe jobs fire. And for those that do inline Dedupe (with a few notable exceptions), the performance suffers tremendously. Also, watch out for Tiering timeframes on EMC gear as well. As I understand it, most of the VMX gear has a an hourly auto-tiering schedule to flash.

 

[NIZMOZ]

EMC here. We have the VNX5300, 5400 and Xtremio in our environment and they are great. The only downers is EMC support is really terrible. Takes forever to get a part. LOL!

I would also recommend looking into TINTRI as they are new on the block but have some nice looking options.

 

[_Gea]

In the end its all about service and SLA.

If you set a price, select Dell/ HP or the preferred SuperMicro Storage Hardware, Intel SSDs or HGST Ultrastar disks with LSI HBAs and a ZFS SAN OS like Oracle Solaris where ZFS comes from or an OpenSource Solaris fork like OmniOS , you will never be able to beat this regarding performance and hardly regarding features.

If you have your own IT department, you can buy enough spare, failover and backup systems to be safe with OpenSource Software and hardware support. Beside that OS support from Oracle for Solaris is affordable (about 1k $ per year) and OmniTi offer support as well for OmniOS.

 

[Blue Fox]

Yeah, no. You pretty much only post to preach ZFS at every opportunity you get. Piecing together your own servers is fine for home use. It would not be appropriate for most companies.

Having proper support is key. You really think paying Oracle $100 a month gets you support? Please stop posting in threads like this until you actually have worked for a decently sized company and understand the sorts of needs that businesses have.

 

[CombatChrisNC]

Well, they could also have a large it department with dozens or even hundreds of people if it's big enough. I wonder if anyone the size of Boeing whiteboxes anything like what he's describing. But you would want a full on team of people just doing these storage setups and need all kinds of spare everythings on hand.

 

[_Gea]

Yeah, no. You pretty much only post to preach ZFS at every opportunity you get. Piecing together your own servers is fine for home use. It would not be appropriate for most companies.

Having proper support is key. You really think paying Oracle $100 a month gets you support? Please stop posting in threads like this until you actually have worked for a decently sized company and understand the sorts of needs that businesses have.

Do you really think that all those SuperMicro server parts for Petabyte storage, massive SSD/NVMe arrays or clusters are for home use? Many use them for professional ZFS storage some with support and SLA and some without. There is a market for high quality storage servers on standard server equippment based on an affordable or free Storage OS.

This must not be the solution for everyone or every problem but can be perfect for others.

 

[Blue Fox]

Plenty of vendors resell Supermicro hardware as integrated packages. You suggested building your own storage servers, which like I said is not appropriate for the majority of businesses. No one cares if the OS is sans licensing costs when it comes to this sort of thing. They don't care what the components are either. They care what the complete package will do for them.

See if you can actually make a constructive and helpful post that isn't an advertisement for ZFS or your side projects involving it? Isn't that why your original account got banned years back anyway?

 

[_Gea]

It seems that you want to replace arguments with personal attacks.

I have no problem if you for example prefer a NetApp over a SuperMicro system from a trusted local vendor or system house with ZFS from Nexenta, OmniOS, Solaris or BSD/ Linux on it. The item is not whether you order and build your own system or buy it ready to use with support for hardware and software. The point is that ZFS system are a true and solid alternative to a Highend Storage in general as they offer quite similar features, mostly a better performance or at higher capacity at a fraction of the costs and they are very stable if you or your vendor follow the "best use" suggestions. The quality and level of service depends on your vendor and your needs and is in not related to ZFS or whatever you prefer. It has a price and that is ok.

I have made my point so you can continue without me.

 

[NIZMOZ]

Sorry, but you will never find an enterprise business buying something like that compared to netapp, EMC, pure storage, Tintri, 3par, compellent and so on. Business wants the warranty, support, onsite response time and they want a business who has been around for years.

 

[DeChache]

Sorry, but you will never find an enterprise business buying something like that compared to netapp, EMC, pure storage, Tintri, 3par, compellent and so on. Business wants the warranty, support, onsite response time and they want a business who has been around for years.

Granted I am in education but we have multiple Petabytes of whitebox ZFS running on I think Open Indiana . We do have Netapp and EMC and they each serve their purpose as well as the ZFS. But I could see us moving our file shares to ZFS.

Times are changing.

 

[NIZMOZ]

They are changing but not towards ZFS.

 

[DeChache]

They are changing but not towards ZFS.

I think you are going to start seeing ZFS come up as a low cost high capacity systems. At least in the research world its gaining ground fast.

 

[Mackintire]

Personally I am a fan of Compellents, easy to use and scale pretty well. And configured right can really perform. But I want to try some of the hyperconverged solutions next. (edit: and with the DELL-EMC merger no one knows yet what will become of Compellent)

As someone with direct knowledge on this:

Compellent has been absorbed into the Dell Porfolio and is undergoing a massive update. Dell already integrated the equalogic iSCSI stack into the SC**** series in the 6.2. They plan on maintaining equalogic for another couple years..but eventually replacing the entire lineup with the SC series with compellent DNA at the core. As someone who purchased the SC4020 over a nimble storage offering, I can say it bring extremely high value for some workloads. I expect both of those solutions to seriously ruin other vender's offerings in the next 5 years. I wouldn't yet call it real enterprise, mainly because of the available support options, but I'm keeping an eye on QNAP's ZFS enterprise offerings.

 

[NIZMOZ]

Also with some knowledge, EMC will be keeping separate within Dell and will be mostly like it has directly towards large enterprise customers while the compellent will continue to take care of small/medium sized businesses.

 

[Mackintire]

Also with some knowledge, EMC will be keeping separate within Dell and will be mostly like it has directly towards large enterprise customers while the compellent will continue to take care of small/medium sized businesses.

SC9000 scales to 3PB in a single SAN array. 6PB if split between the NAS 8600 head and the SC9000 SAN head. Keep in mind you can also use multiple arrays in one federated logical interface. At the moment the SC series is good for certain scenarios, so don't treat it as the end all be all.

 

[NIZMOZ]

SC9000 scales to 3PB in a single SAN array. 6PB if split between the NAS 8600 head and the SC9000 SAN head. Keep in mind you can also use multiple arrays in one federated logical interface. At the moment the SC series is good for certain scenarios, so don't treat it as the end all be all.

That isn't my words, that is from Dell and EMC that we met with 2 weeks ago. when they were describing the changes in the company with EMC. The Compellent isn't designed for large enterprise in mind. EMC always has been.

 

[Mackintire]

That isn't my words, that is from Dell and EMC that we met with 2 weeks ago. when they were describing the changes in the company with EMC. The Compellent isn't designed for large enterprise in mind. EMC always has been.

That I agree with, and sounds similar to the discussions we had with Dell.

 

[Child of Wonder]

Disagree. I have Pure, and I'm not impressed. My sales guys and TC are shady as all hell. I've caught them lying to me about the capabilities of their product 4 times.

Damn, that's a lot of FUD in one post.

A sales team certainly could miss when it comes to estimating data reduction. But any prospective customer can get a Right Size Guarantee from Pure before buying that will guarantee the array will hold X TB based on the data the customer plans to put on it. Tell the sales team, for example, you have 10TB of SQL, 10TB of VDI, 10TB of Exchange, and 20TB of general virtual server data and they'll create a guarantee that estimates the data reduction you'll get which will turn into an effective capacity the array will provide. If the array fails to hit that, the customer gets a capacity expansion for free.

Over exaggerating Dedup and Compression, Not being fully transparent on the post compression process, not being fully transparent on when that process is shut off because the array is over loaded, the documentation on their own website, oracle, and VMware saying something completely different then what they say in meetings.

The post compression process is simply this: we run in-line compression on the data as it comes in, then use spare CPU cycles to run a deeper level of compression on the data later on. Deduplication behaves similarly. Data reduction is never "turned off."

Migrating from EMC powerpath to Pure on a Linux host means a Linux host reload but they will tell you something different.

How is migrating from EMC Powerpath on Linux to Pure Storage, which doesn't need Powerpath, a Pure Storage issue? If you need to reload your iPhone every time you pair it with a different car, is that a car manufacturer problem or an Apple problem?

No local user authentication other then a single pureuser account? whaaaaat?

Yep, only one local user account. Sync up the array with Active Directory or LDAP, set up RBAC, and then change that one local account's password to something obscure, save it somewhere, and use your own account. How would managing several local user accounts make your life any easier?

The boxes I demo'd were basically rebranded Dell boxes running Centos. The next gen hardware was nicer and more modular, but still based on centos.

Old hardware was OEM'd but the //m series released in summer 2015 is our own custom hardware. How was that any different than several other vendors in their first few years of existence? Our arrays do not run CentOS.

Further more, the sales team will underspecify what you need, so when it gets on your floor and you ram it into the ground, you'll have to upgrade it.

Ask for the Right Size Guarantee and if the sales team undersizes you get a free upgrade. Simple. If you don't want the array within the first 30 days of owning it, send it back for any reason for a full refund. Simple.

Granted swapping engines is easy and should be seamless, but you are taking down half your storage compute power while you do that.

Swapping engines IS easy and IS seamless. Yes, you will be running on only one controller for a combined ~20 minutes during a controller upgrade but performance remains at 100% during and exceeds 100% after. Is this better or worse than the overwhelming majority of storage vendors? Customers also get these controller upgrades as part of their maintenance renewal every 3 years (maintenance that will only ever increase in price from the initial purchase if the customer buys more capacity).

Pure will tell you, plenty of people are running away from EMC to go to Pure.. But what Pure doesn't tell you is how many people are running away from Pure to go back to EMC.

I am not aware of any customers in my three state region that have purchased Pure and gone back to EMC. Can you share any with me?

Oh and the kicker? They will murder the entry level price to get it on your floor.. But when it comes time for upgrades, they really stick it to you.

Are you asserting that other vendors will maintain aggressive initial purchase discount pricing for as long as a customer owns an array? Pure may have to get aggressive, like anyone else, to win an initial sale but when the customer gets free controller upgrades as part of their 3 year maintenance cycle which eliminates forklift upgrades every 3-5 years, can evacuate existing shelves of SSD so they can trade them in for denser shelves (you don't rebuy TB you already own), has Pure support proactively monitoring the array and performing software upgrades for them, and can seamlessly migrate from one architecture to the next (moving to NVMe SSDs and shelves someday, for example) without downtime or data migration, customers overwhelmingly see the savings in CAPEX and OPEX going Pure.

If you compare Pure with XtremIO in terms of technology, hardware, layout, design, support, management, upgradability and TCO? Xtremio walks all over Pure all day everyday. If you look at Pure, take a second look at XtremIO.. Because while the cost to get into XtremIO maybe more expensive, the total cost of owning it will be about the same as Pure in 4 years.

We have compared Pure to XtremIO many times (don't really run into it much lately):

- Technology and hardware? Why would a customer want expensive and low density eMLC drives vs today's cheaper and more dense 3D TLC when a storage array can get the same performance and longevity, including warrantying the 3D TLC drives for as long as you own the array without increasing maintenance costs?

- Layout and design? Have you seen the back of an 4+ cluster XtremIO implementation vs Pure? Talk about a complex, 23RU+ cable nightmare.

- Support and management? Everyone who loves EMC support raise their hand. Everyone who loves Java based management that requires its own separate server raise their hand.

- Upgradeability? Why would I want to buy a 10, 20, or 40TB node only to lock myself in at that node size, only able to expand my array at that increment in pairs of two until I hit 8 nodes? Why wouldn't I want to be able to grow my array in different increments or even non-disruptively trade in low density shelves I already own for bigger ones? Why would I want to migrate all my data off an array for a major software upgrade? Why would I want to pay for professional services for an upgrade? How does any of that benefit me and my business?

Introducing the Newly Expanded Evergreen Storage – Guaranteed Capacity That Stays Modern and Dense

- TCO? Buy technology that requires more rack space, more power, more cooling, more cables, more complexity that I'll have to forklift upgrade every 3-5 years or if I want to move to higher densities VS buying technology that has maintenance costs that don't go up unless I buy more capacity and include controller upgrades every 3 years and the ability to move to denser shelves or newer flash technology non-disruptively, and protecting my initial investment by getting trade-in if I have to move up to a faster controller or add denser shelves.

Oh.. and with questions on wether or not Pure is actually profitable as a company is a huge red flag.

<link>

Dunno about you, but that's not good.. But the rumor mill says while they are a public company, they a looking to be swallowed up by a Cisco for UCS, or Oracle for exadata.

Pure isn't profitable today. We just released our Flashblade product, an all flash, total ground up software and hardware build, for file and object storage, which has been in development for years. As a publicly traded company anyone can look at our financials and see where the money is going. Up to others to decide if we're spending money appropriately as an all-flash company as the entire world begins to move away from spinning disk and towards flash media.

As for the rumor mill, I won't comment one way or the other but I can say is it's pretty ironic when companies that they themselves are being bought make negative claims and predictions about a competitor being bought out. Glass houses and all. As a Pure employee I'm bullish on what the company has done and will be doing. Rumor mills are just noise and distraction.

 

[Shockey]

Why would I want to migrate all my data off an array for a major software upgrade? Why would I want to pay for professional services for an upgrade?

First statement isn't true anymore and all upgrades both software and hardware are done online with the released of 4.x family of code.

All XtremIO code upgrades are covered under support contracts.

 

[lordsegan]

It is not right for everyone, but I've had some positive stuff going on with OSNexus.

 

[NIZMOZ]

First statement isn't true anymore and all upgrades both software and hardware are done online with the released of 4.x family of code.

All XtremIO code upgrades are covered under support contracts.

Correct. Only going from 2.x to 3.x required a destructive upgrade. 3.x and up it is an online upgrade. We have had no issues doing it either. Plus you are right, Support does it al.

 

[Langly]

Goodness this thread is neat. Just saw this now and wanted to get 2 cents in for someone that the numbers+performance game

OP, there are a lot of great storage options out there. Definitely look at them all if you have time and find what looks best. Take your partnership with the sales/VAR serious too. If you aren't comfortable with your sales group and/or VAR then shop around. I still use the same Cisco VAR from company to company because I have 100% confidence in them. Same thing with my old Netapp VAR if I need them. I've got Nimble/Pure/EMC/Netapp in varied areas that even the older products are great.

All flash arrays are coming way down in price due to tech behind the disks becoming more affordable. I've had all the big vendors knocking on the door trying to sell things which I'm getting sorely tempted to for some of my database needs. I'm liking the new Netapp features and the flexibility/ease of access from the EMC Unity line. Although the Unity POC had some ESRS setup issues due to items not being documented which is frustrating. New product line woes.

Child of Wonder, that line by line ripping apart of a Pure customer's feedback and experience is just mind boggling when honestly it seems that you are a Pure employee or partner. Pure has a good product with good pricing but the numbers game makes me second guess buying them or Nimble until the product is bought out by a larger company to keep things afloat. Don't do the company a disservice with venting online when you can just let the product speak for itself. I have a lot of buddies/old colleagues that work at Pure, there are definitely shady as hell sales guys there. ALL vendors have shady as hell sales guys.
In regards to the feature sets on the box, it is a great product for its price point. With how far the XIO line has come since the 2.0 to 3.0 laughingstock of an upgrade, I'd still give them a look although not want to pay the EMC tax if I don't have to. That ripping apart though really is stopping me in my tracks on the Pure point. We have a Pure array and we love it in our DC, my wifes hospital chain she does telecom for just bought quite a few Pure boxes for raw power. No need to rip people apart for their opinion and experience.

OP again, storage hardware is MAJORLY competitive right now. The market place is shifting heavily and you will keep seeing great changes from vendors. Vendors right now cannot afford to be stagnant otherwise they will get eaten up by all the great competition out there. Take the time to find what will best fit your company and get some POC's going. Sales guys will be able to size up your IOPS and sizing needs accordingly. Make sure to plan all the feature sets you will need and if your IOPS needs aren't crucial, don't be afraid to go cheaper on a normal traditional array or a hybrid-AFA setup. I can't wait to get spinining disks out of my datacenter in the next 5 years. 16tb EFDs coming down from vendors now is exciting!

 

[Child of Wonder]

First statement isn't true anymore and all upgrades both software and hardware are done online with the released of 4.x family of code.

All XtremIO code upgrades are covered under support contracts.

Correct. Only going from 2.x to 3.x required a destructive upgrade. 3.x and up it is an online upgrade. We have had no issues doing it either. Plus you are right, Support does it al.

Glad this has changed. Are professional services still required for hardware upgrades? Going from a single node to a cluster or expanding a cluster?