Weak AF array sales at NetApp leave analysts feeling cold NetApp reported quarterly revenues that were weaker than expected last night – reduced by feeble sales across flash arrays, OEM and EMEA. Execs fingered poor execution on renewals and foot-dragging by data centre buyers. The storage firm's shares dipped more than 6 per cent as it reported results (PDF) for its fourth quarter …
Ok, when NetApp rose, it was because companies overconsolidated and overwasted. Not only that, but Microsoft, VMware and OpenStack lacked built in storage solutions. Most storage sales were measured on the scale of a few terabytes at most. Consider that a 2TB FAS 2500 series cost a company $10000 or more using spinning disks.
Most companies ran their own data centers and consolidated all their services into as few servers as possible. They went from running 5-10 separate servers (AD, Exchange, SQL, their business app...) costing $2000 each to 3-10 VMware servers costing $5000 each plus a SAN and an additional $2000+ in software licenses each... to run the same things.
Performance dropped considerably when they made that shift. Sure, they were supposedly easier to manage, but the management began to realize that systems that used to take 1 average skilled employee and 1 consultant to manage now took a team of full time employees and a lot more consultants to run.
Performance was almost always a problem because of storage. NetApp made a fortune because they could deliver a SAN which was relatively easy to manage that could handle most small businesses data.
What got really weird is when the bosses wondered how they went from $100,000 IT costs per year (people too) to $500,000 or more and no matter how much they spent on tools to make it more reliable and more robust, they always found themselves with the same outages and increasing costs.
Enter the cloud.
Companies could move their identity, mail, sharepoint, collaboration and office tools online using a relatively easy migration tool which took a few days to weeks.
SQL and their company app could be uploaded as VMs initially with little effort and with some effort, they could move their SQL to Azure’s SQL.
Now, they can downsize to one IT person and drop their costs to about $100K a year again.
The catch is, since we no longer need a dozen IT guys and consultants, no one left knows what either NetApp or Cisco is and they’re just using the simple pointy clicks UI to do everything. Their internal data center is being spun down and finding its way to eBay instead.
Then there’s whoever is left. They find that by replacing their servers with new servers containing disks, they can use VSAN, Storage Spaces Direct or Swift and not have to spend money on external storage which actually has a lower aggregate performance and substantially higher cost. Not only that, but they’re integrated into the systems they run on.
NetApp has no meaning for cloud vendors because MS, Google, Amazon, Facebook, Oracle can all make their own. In some cases, they even make their own hardware.
NetApp will still have a market for a while, but they will become less interesting as more services are moved to the cloud. After all, most companies depending on NetApp today probably have just enough performance to continue operations and as more systems go to the cloud, they’ll need less performance, not more.
There will be organizations like military and banks who will still need storage. And of course there are surveillance systems that require keeping video for 2-10 years depending on country. But I believe increasingly they will be able to move to more cost efficient solutions.
NetApp... I loved you, but like many others, I have now spun down 5 major NetApp installations and moved either to cloud or to OpenStack with Ceph. My company is currently spinning down another 12 major (service provider scale) NetApp solutions because we just don’t need it anymore.
I wish you luck and hope you convince HPe to buy you out like they do to everyone else in your position.
All true, but if you look carefully at NetApp's investments in Data Fabric and particularly at NKS you'll see a different picture of NetApp emerging. It is all about cloud first strategies and getting data to the right place for the right cost. Dell, Pure and HPE are not playing in that space. VMware on AWS is a fraud that exploits enterprises spending even more money on Dell rather than making a meaningful transition to cloud.
Honestly, I think the NKS platform looks ok, but I expect that it's only a matter of time before all three clouds have their own legitimate competitors for it.
Don't get me wrong, I'm not saying it to be a jerk... as I said, it looks ok. But it's obvious progression for K8S, I've been building the same thing for internal use on top of Ceph at work. I'm pretty sure anyone trying to run a fault tolerant K8S cloud is all doing the same. But to be honest, if you're doing K8S, you should be using document/object storage and not volume storage.
If you're running Mongo or Couch in containers, I suppose volume or file storage would be a good thing. But when you're doing "web scale applications" you really should avoid file and volume storage as much as possible.
I just don't expect NetApp to be able to compete in this market when Microsoft and Amazon decide to build a competing product and pretty much just toss it in with their existing K8S solutions.
You give the example of the $100,000 spend now turning into $500,000 spend. Perhaps true, perhaps not, I think it depends on the customer. I have seen, met and done business with probably close to a thousand at this point in 15+yrs in my career.
What I'm seeing in large 10,000 employee+ shops that thought, the cloud was grand, it's not. Or I should say, with poor planning it's not. DBA's taking server hardware to AWS in a 1:1 fashion, and getting billed tens of thousands of dollars per month; executives quickly realize, cost savings are only realized with a transformation of infrastructure, policy, mindsets, and people.
That CIO or VP that said, "we're going 100% cloud! And we're doing it in 18 months." Is no longer employed there (Repeated in many orgs I've worked with.). For as many cloud migrations take place, I suspect, you'll also see quite a bit of reverse directions, or hybrid models.
These days, it's not cloud-first. It's more cloud smart, or cloud ready.
This post has been deleted by its author
You will see almost all companies going hybrid but the cloud reverse claim, in my experience, is largely an urban myth. Many, even large orgs, false started on their cloud journey but are focused on getting it right rather than reversing it. Almost all orgs are in agreement that hybrid will prevail but there is still a cloud first mentality, and it's gathering pace.
I don't disagree on many points. I've seen some pretty botched cloud gambits.And those are almost always on the companies that go to the cloud by copying up their VMs as quickly as possible. It's like "If you actually need VMware in the cloud... you really did it wrong"
The beauty of the change is that as we systems that genuinely belong in the cloud... like e-mail and collaboration is going there as SaaS and it's working GREAT. Security for email and collaboration can't ever work without mass economy and 24/7 attention from companies who actually know what they're doing...not like Cisco AMP or ESA crap.
A lot of other systems are going SaaS as well... for example Salesforce, SAP, etc... these systems should almost, by law have to be transferred to the cloud if for no other reason than it guarantees paper trails (figuratively speaking) of all business transactions that can be audited and subpoenaed. Though that's true for email and collab.
Systems which are company specific, they can come back home, and then eventually over time get ported to newer PaaS type systems which can be effectively cloud hosted.
I actually live in terror of the term "Full Stack Developer" since these days it often means "We don't actually want to pay for a DBA, we'd rather just overpay Amazon"
...He also said Netapp was "really betting on the hypothesis that enterprise-grade hybrid cloud infrastructure will displace first generation hyperconverged".
The public cloud data services business is much smaller: "We expect that the general availability of Azure NetApp files... should be a step function increase in our opportunity with cloud data services."...
Can anyone help me with this, anyone?
Azure NetApp files is Enterprise NAS running on NetApp HW within Azure and being resold only by MS. It is cloud native and not even sold by NetApp directly. That is a huge new upside, think of all the MS sales reps and new opportunities it will bring. Same thing with GCP, they are doing the same and in both instances you pay the hyperscaler and not NetApp. NAS at scale with performance is too hard for them to build themselves...That is why they resell NetApp as an OEM cloud partner.
I dont think on prem storage is dead just yet. Cloud storage is still pretty expensive for large workloads. We just selected Pure over netapp to replace our out of warranty 3par fleet. HP support has gone straight into the toilet, so we needed something to replace for on prem storage. Between the two, netapp was completely confused as to what they were trying to sell (storage or cloud?) and all of their interfaces seemed dated and disjointed. With the dedup and compression available with Pure and NVME, we are going from around 1pb raw to around 300gb raw nvme.
We are going from 7 racks of HP spinning rust to 12u of pure nvme. Amazing, so there is some innovation happening on the storage space.
Cloud isnt going to be a fix for large organizations until they are willing to rewrite and retool to use native cloud tools instead of IaaS. The cost savings are there, but not for lift and shift of existing vm's from on-prem to cloud.
The fact you have gone 12U and 300GB raw is just the change in technology. Compression, deduplication and SSD/NVME all contribute to the reduction in size. As with any vendor, the grass is usually greener in the other field as new kit does not have as many failures. 7 racks for a PB is very low density/old kit so high failure rates and lots of support calls should be expected.
"Cloud" does not replace storage, just the billing model. And 186,000 miles a second isn't just a good idea, it's the law. Low latency on prem storage will be with us as long as legacy applications exist. It's possible to do all sorts of things in "The Cloud" but not everything.
I don't disagree. I still see the occasional UltraSparc5, AS/400 and Window NT 4 machines in production. Legacy will always exist... but I think you're overestimating the need for low-latency on premise storage.
As latency to the clouds are decreasing and bandwidth is increasing and availability is actually often rivaling on-premise location isn't the hot topic.
We used low-latency storage for things like fiber channel because we were oversubscribing everything. But if you consider that massive banks still run on systems like IBM Z which seem really amazing, but performance-wise are generally obscenely over-provisioned. A well written system can handle millions of customer transactions per day on equipment no more powerful than a Raspberry Pi... and they did for decades... on horribly slow storage.
The question is... what do you really plan to run back home anymore? Most of the reasons you've needed extremely high end storage systems in the past have moved to the cloud where they logically belong. This means that most of what you're running back home is actually non-business systems anymore.
A major company will probably have something like an in-house SAP style system and a bunch of other things like file server which no one uses anymore. Everything else will be moved to the cloud with or against IT's "better judgement". Remember, you don't need the IT guy to sign up for Slack, the boss does that with his own credit card while sitting in a meeting.
The cloud doesn't replace storage... it replaces the systems using storage.
Now... let's assume you're working for a news paper or a television station where you need local storage because 1000+ photos at 20megapixel RAW or 25 hours of video at 12Gbp/s needs to be stored somewhere. These days, you pay a lot of money for your storage, but you also have a choice of easily 10 legitimate vendors and maybe another 200 "won't make it through another funding round" vendors. Right now, there's lots of choices and all those vendors still have lots of sales keeping them in the black.
Now, as more and more services are migrated to the cloud. The storage systems at most companies with more "plain vanilla" needs will free up capacity on their local storage. If they refresh their servers again, they'll choose a hyperconverged solution for the next generation.
This will mean that the larger storage companies will dissolve or converge. If they dissolve, they're gone. If they converge, they'll reduce redundant products and deprecate what you already have.
As this happens, the companies with those BIG low latency storage needs will no longer be buying a commodity product but instead a specialty product. Prices will increase and the effected customers will be substantially more conservative about their refresh cycle in the future.
Storage is ending... sure... there will always be a need for it in special cases, but I think it will be a LONG time before the stock market goes storage crazy again. And I don't think Netapp, a storage only company will survive it. EMC is part of Dell and 3Par is part of HP etc... companies which sell storage to support their core business. But Netapp sells storage and only storage, so they and pure will be hurt hardest and earliest.
trying one last time, double posted, hopefully cleaned up now...
quote What I'm seeing in large 10,000 employee+ shops that thought, the cloud was grand, it's not. Or I should say, with poor planning it's not. DBA's taking server hardware to AWS in a 1:1 fashion, and getting billed tens of thousands of dollars per month; executives quickly realize, cost savings are only realized with a transformation of infrastructure, policy, mindsets, and people. /quote
Exactly what we are seeing. Just calculated a price to put a mid size DB on RDS, $1200 a month. AWS has been good for stuff like authentication, DNS, backup targets etc, but for large loads, we are doing it much more economically on-prem
Cloud CAN be cheaper, but as you said, you have to use 'their' tools and services, not lift and shift VM's from on-prem to cloud.
BTW> On prem storage is still alive, just purchased Pure after a bakeoff with Netapp. We are replacing 6 HP 3par's due to maintenance costs and the fact that HP support has gone directly into the toilet. 6 racks of spinning rust reduced to 12u. Crazy.
Netapp is still using very dated interfaces, does not use a consolidated management web site or console, and seems to be very confused as to whether or not they are cloud vendor or a storage vendor. The sales teams are focused on top-down sales which is a negative in my book as well.
Netapp was much cheaper, but at least netapp was cheap, pretty much the only real pro we could find with them. Not surprising that their sales are not great based on our recent experience.
There is a truth out there. The idea flash "wears out" while hard drives do not may be true, but the time it takes flash to wear out is much longer than many anticipated. Long enough that the overwrite wear of flash actually takes longer in many cases than the physical wear out of hard drives (the upward curve of the HDD lifetime reliability "bathtub"). As a result, all-flash arrays can be amortized by customers over a much longer life cycle, meaning fewer refresh cycles.
This is hitting every storage array vendor who made a mint over the last 4 or 5 years replacing spinning disks with SSD based arrays.
Add to that over the last 5 years the number of organizations who have moved things like email to the cloud. There was a time an organization would dedicate an array to Microsoft Exchange. Not any more.
The overall storage array market is declining, and the refresh cycle is stretching out to 5 or 6 years. This means a reset of expectations, and a need to expand beyond traditional solution areas. That is a big contributor to the flattening growth curve of all-flash systems.
Funny (?) story: I once had a very odd conversation with Thomas Kurian, whom I knew from Oracle days, at a conference. He was uncomfortable and distant throughout, even though I thought we had been professionally close before.
It was several weeks before I realized I had been speaking with George, whom I had never met before.
we're using the netapp all flash units as a 'gateway' for onsite cifs/nfs, tiered to on-prem S3 and also to Azure/Amazon.
The interface seems bang up-to-date to us and we are also controlling it with ansible playbooks.
it's the trojan horse allowing us to move away from a certain other major storage vendors, and their mid-tier was very unmanageable
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
