Don't take this the wrong way, Pure Storage – are you the next NetApp? Back in 1992, NetApp was founded and competed with Auspex in the file storage array market. It overtook Auspex and grew and grew, entering the Fortune 500, becoming a storage platform company and a multi-billion dollar revenue corporation. No other startup has managed that feat since. Could Pure Storage be the first one to …
Mr Coward ......
There are many tabs at the top of el Reg, the one you are in is Storage, it is where Storage related articles are posted.
Should you wish to see other non-Storage related articles, you could pick another category - but then I guess you will get lots of posts about that too. {sigh}
el reg signed up a bunch of bloggers to post more articles on the site. They reached out to me about 12-18 months ago but I wasn't interested(lost interest in writing such articles not much in tech inspires me these days, I used to blog more in a week than I have in the past year now). I wrote a few hundred articles many in depth on my personal blog over a 3-4 year time span.
Trevor is another one of the folks they signed up, though not sure if he was a blogger or not.
I honestly don't understand the big deal about whether Pure is culturally like NetApp or EMC.
I've been at Pure for 2.5 years and previously worked for NetApp for a very long time - nearly 13 years in total. I have never worked at EMC so I can't comment on their culture. I loved working for NetApp but I saw what Pure wanted to achieve and I wanted to be part of that more.
Yes, I work with a lot of ex-EMC people - but guess what! They are nice guys - they don't have two horns growing out of their heads and they don't speak with forked tongues! :)
Every storage vendor is full of people who worked for other storage vendors in previous lives, and by default a lot of them will be ex-EMC and a lot of people who worked for other vendors are now at EMC.
What's important is that people work at a company that they enjoy working for and passionately believe in what they are doing and the difference that what they do will make for their customers. That's why I'm at Pure and that's why all my colleagues are here too - regardless of where we came from previously - and it helps that we like Orange a lot...
I've got something. A bit dated from 14. ICD Report courtesy of EMC. This has EMC, Pure, IBM in one, two and three positions respectively. Interestingly IBM leads, by a lot, in terms of total capacity shipped, but EMC leads in revenues. Not sure the point EMC wanted to make with this is that IBM is giving people better deals on storage.
http://www.emc.com/collateral/analyst-reports/worldwide-all-flash-hybrid-flash-array-forcast-vendor-shares.pdf
I think what stands out about Netapp is they found a niche that was very important to many (i.e. NAS) and they filled it. They also stayed well ahead of the competition in this area for many years to the extent that if you wanted a NAS enterprise appliance, you generally went to Netapp.
Others have caught up since, to an extent, and also offer decent block storage, which is something Netapp has struggled with over the years, certainly in relation to its NAS offering.
The market has changed a lot: smaller consumers who would previously buy entry-level or mid-range storage are moving more to the cloud or to outsourcers. Of course, storage is still required but this tends to be on more enterprise level tiered storage or more recently, the hybrids. There's also the DIY Ceph and ZFS offerings and of course, the outsourcers and clouds can utilise AFAs, whereas the average customer will find that such performance is only needed for a few workloads and they're more interested in reliability and cost for the rest.
So everyone's chasing the big enterprises who still have stuff in-house and those who are providing for the rest. And there's a lot of people offering similar products in the same markets. The AFAs are all very much the same and for them it's all about performance. The tiered stuff and traditional arrays are all legacy and they can afford to add features to solid, yet old-fashioned platforms, but they're always going to be inefficient due to their legacy designs. The hybrids and software guys are the interesting ones: trying to put something new into an already saturated market isn't easy so it's the differentiators that count. Netapp had a huge, albeit now diminished differentiator. I'm not that sure that Pure has.
@AC ''AFAs are all very much the same". I've tried out a few and I can promise you while the sales pitches often sound very similar put them to an actual test both the arrays and the support is very different. I would very strongly suggest doing a PoC (proof of concept). The AFA vendors have hired some of the best sales people out there - don't just take their word for it, test it for real. Trusting an AFA sales person can be a CV generating event ...
@AC #2 - Yes I agree with you. My point was that they all are offering the same thing - performance mostly. The reality is different, as you say. Performance isn't always that great and it's never going to be better than buying a big box with a couple of TB of RAM.
Despite what they tell you, AFAs are a niche market. A very lucrative niche market and if you have applications which demand low latency and a hell of a lot of read cache misses then they should be towards the top of your shopping list.
For most people, a few applications, if any, demand ultra-low latency and what's more important is getting proper support and making sure the thing doesn't fall over in the first place. If an enterprise array can't claim at least 6x9s then why the hell would you consider it?
I don't think anyone should ever buy something new without doing a PoC. And test the support process during that PoC too. And don't let the vendor set the agenda.
AC #1
I love the fact that the writer consider Whiptail a success story when Cisco doesn't even sell the product anymore.
But more importantly, it bugs me how biased this is towards Pure like they are doing something brand new and fresh. Yes, maybe they were 3 years ago but now it just seems like they are trying to come up with stuff, even if its over a year out to stay competitive. I mean really, if you are going to sell Pure in this article, don't sell me on futures. Sell me something now...
Yawn, yawn, yawn..
Pure is going to skip all of the middle part of the NetApp story and get right to the dying and going out of business part. There is only one possible exit. They will be acquired. Pure made a small group of people a lot of money, but ultimately the music will stop and investors will have no where to sit.
Increasingly, the people who care enough about storage arrays to ferret out the nuances between boxes will be told to go do something productive for the companies they work for. The world is changing.
Focusing on storage arrays is like race car drivers worrying about spark plugs.
This post has been deleted by its author
Wherever they're placed on the market share league, they did it in four years of being GA, with a sales team and channel a fraction of the size of the big guns, in fewer countries. At current rate of growth, they'll be as big as NetApp in half the time. Say what you like, but customers like them.
Last year's Gartner market share analysis - SSDs and Solid State Arrays said:
"SSA market sales leapt to $1.299 billion (up 120.3%); EMC gained considerable ground, due to a full year in the market, and led with a 34.1% share, followed by IBM at 18.0% and Pure Storage at 11.5%."
Since then IDC has done quarterly marketshare reports and in 2015 Pure trailed behind EMC, IBM and HP respectively. It has swapped places in share with HP and IBM a few times in '15 and is closest to HP in a near tie for 4th place.
For a while it looked as if storage controller based features were a good story and customers bought the swiss army knife controllers with "snap" and "dedupe" bells and whistles no matter how unpredictable these system where. I'm talking about Ontap in particular.
Eventually customers saw through the BS and refused to pay lots of $$$ for a Storage Controller with a CPU they could get a a fraction of the price in a commodity server.
Hypervisor/Converged Vendors have caught up and all the software "features" are now available in the HyperVisor/OS.
So we're back to the point where the Server tells the Storage what to do. And the storage better do it fast, which is why Flash is so popular.
Eventually we'll be back at block protocols - at least where performance and reliability matters.
Maybe the big Hyperscalers, Google or Facebook will come up with a home-brew storage OS talking to dumb JBODS, but the days of the traditional tier 2 storage vendors are over.
Block? You still want to emulate heads and cylinders on top of flash that doesn't even have this anymore? Along with a dinosaur SCSI protocol?
In case you haven't noticed, the storage world is shifting towards cloud and software defined. Not a very good place for Fibre Channel with is extreme hardware dependencies, where drivers & firmwares all need to be carefully aligned to be supported because of the "robust" protocol. *lol*
Yes, you might use iSCSI as a block protocol in ethernet-based cloud and software-defined environments. But why?
NAS has become popular for performance and reliability over the last decade or so. See NFS Datastores, Oracle dNFS, Microsoft SMB3 etc.
Object might become the new block but that remains to be seen. Most new applications already use S3.
Then please tell me what the cloud runs on ? Hot air ?
They use block in the datacenter - and who cares what the cloud protocol flavour of the months is ?
Up until recently backend storage shelves were connected by FC, now SATA - all running the SCSI protocol.
"Software Defined" happens in the front end where the consumers are.
It's the same for Software defined networking - Routers and Switches still move packets.
Remember that the cloud is a method for packaging and pricing Services. Somewhere(in the cloud) there is still technology that actually has to deliver the services.
Object is and addressing scheme - some abstraction layers , some networking protocols - all stored on block storage. Maybe heads and cylinders will be replaced by something else, but the closer you can get to the hardware the better.
Software Defined and Object - actually introduces an abstraction layer between storage consumer and storage hardware.
Maybe one day you can have software defined magnetic platters or nand chips :)
Not one storage startup since 1992 has made it to standalone, profitable, platform status; that's a 26-year period.
Is this a true statement, I might be reading it wrong, but Isilon was profitable before EMC purchased them, Promise is standalone profitable. NexSan and Dot Hill both were before they got acquired as well.
This whole story just seems like a retread of a blog some NetApp marketing guy wrote before he got laid off last year. His blog has since been shut down, but some of the statements seem very familiar.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
