NetApp to 'realign', cut 600 jobs worldwide NetApp has announced it will cut 600 jobs around the world, as part of a plan to adopt “... a business realignment plan designed to focus its resources on key strategic initiatives and streamline its business in light of the constrained IT spending environment.” The company announced the plan today in a form 8-K filed with the …
Well there's object storage, nobody cares about that. And that cloud based file sharing bandwagon everyone jumped on isn't really a core technology unless NetApp actually does want to become a software company as threatened.
But really they only have two product lines, there's not a lot of fat to trim. Sounds like maybe they'll be sacrificing a few salespeople and PSE in small offices, middle managers, support staff, devs on products that don't have any major work going on, stuff like that.
...their age is starting to show. A few years back it was great stuff, the performance wasn't always the best but it was a solid Swiss Army knife product, you really could do just about anything you wanted on FAS storage arrays and DOT. I had a few of them myself and aside from some bad sizing which I inherited from prior to my joining to company, they were quite good.
The problem I saw was with their hybrid story at the time. Everyone else was doing tiering of some sort, EMC, Compellent and 3PAR (pre-acquisitions), Hitachi, IBM all had a tiering strategy to accelerate reads and writes. This meant that midrange and up storage buyers could use their existing arrays and start adding in the performance advantage of flash without paying huge costs, the array intelligence would move data to where it needed to be. NetApp offered Flash Cache cards (previously PAM cards) but they only accelerated reads. The only way to shoehorn flash into a FAS for write improvements was to put in all-flash aggregates and manually move workloads between them.
That meant bigger costs and manual effort compared to the competition. I hate both of those two things. I had a friend in another organization with a VNX5300 add 2 drives for FAST caching when his array started slowing noticeably and the performance improvement over his 50 spindles of disk was like 100% with latencies getting cut way down. 2 SSDs that popped right in to free slots in their existing enclosures. That's not a big cost in the grand scheme of things. I remember the discussion I had with my then-director about if we could do something similar (pre-Flash Pool) and the answer had a 6-digit price tag.
They needed to innovate more (and more often), their "no one wants tiering" strategy was a huge failure and now their products are trying to keep up with what everyone else has had for years.
"They needed to innovate more (and more often), their "no one wants tiering" strategy was a huge failure and now their products are trying to keep up with what everyone else has had for years."
Actually, no, auto tearing where you physically copy data from one class of disk to another is what has been found is a huge failure. Ask any customer who has a database that gets hammered once a month only to get autotiered down to SATA disk and is in the worst possible place when it comes time to get hammered again. Physical Auto Tiering is a day late and a dollar short moving data after it is too late. Flash Cache and Flash Pools accelerate workloads immediately, not after it is too late.
What technology is there that everyone has had for years that NetApp Doesn't have anyways?
Not sure how you figure "they're currently hemorrhaging staff" They are letting do of 600 and there are 274 postings on their job board right now so a net of 326 positions reduced for a company with 13,000 employees is 2.5%. Hardly a hemorrhage. HP cutting 34,000 against 314,000, slightly more than 10%, now that is a hemorrhage.
Also not sure what you mean by "BTW are we doing the old lets intentionally confuse the capabilities of 7 mode with those of cluster mode again?" both 7 mode and cluster mode do virtual auto tiering, so not sure what you mean by that.
For a stagnating company publicly offloading staff, they have a lot of jobs on offer. Based on the people I've spoken to I'd bet the vast majority of those roles aren't net new. They're actually there to replace losses in the ongoing exodus. Netapp's inability to innovate over the last 4 years and its decision to ham string cluster mode with all the underlying baggage of the FAS architecture is coming home to roost.
I don't agree that the 'no tiering' strategy caused their current issues. The main problem, imho, is CDOT.
They've spent so long trying to bring this to market and make it 'a thing' that the industry has moved on. Now they're left to monetise years of development and convince their customers that they need to switch. I'm guessing that's not been a success.
I don't get how anyone posts that NetApp doesn't have tiering. NetApp has had tiering for 7-years and hybrid arrays for almost 3. Just because NetApp does it differently where the data is instantaneously copied, not moved, to faster capacity for immediate acceleration vs. waiting until the next day when it is too late, doesn't mean they don't have it.
Regarding the problem being CDOT, not sure where that comes from either. Agreed the transition from 7-mode to CDOT is not easy, but I have not come across any storage admin who says Nah, I don't care to have a system that I can scale up, scale out, retire controllers, tech refresh, swap out entire disk shelves etc without ever having to take an outage. Nope, no use for technology like that.
That isn't tiering it's called caching and everyone has had it since the year dot. Netapp had to use that method because tiering wasn't possible due their underlying file system WAFL, hence the need for a stopgap with PAM / flash cache until they could get hybrid aggregates out the door.
Even then all tiering remains per aggregate per controller, it's not system wide and PAM has it's own issues being a simple non mirrored read cache. Not to mention what's possible on 7 mode isn't always available in CDOT cluster mode,.but intentionally mixing the two's capabilities to confuse seems to be par for the course these days.
Also please lay off the CDOT marketing, yes possibly it's better than 7-mode at scale, but really it has to be the most clunky and least elegant clustered storage solution on the market. 10 years of development to effectively cobble together a pseudo cluster made up of traditional FAS clusters so you can effectively scale, what next ?
Interesting statement because if you read the description of EMC Fast Cache, they define it as a "tier" as does NetApp with Flash Cache and Flash Pools. You may say "That isn't tiering it's called caching" , but the two largest storage companies in the world, EMC and NetApp define SSD cache as a "Tier"
In the Tech Target article below, written by the President and Founder of Storage Switzerland, he discusses caching with SSD's and Flash. In the very first sentence it states "Caching is used to automatically move or copy the most active data to a higher-performing tier from a slower one." and goes on to use the word tier 13 more times to explain this technology.
http://searchsolidstatestorage.techtarget.com/answer/Comparing-write-through-write-back-and-write-around-caching
Net is the old definition of tiering where one physically moves data from one class of disk to the next is being redefined by instantaneous SSD and Flash Virtual tiering again which NetApp has had for 7 years. It's not just NetApp saying that, but EMC as well.
Regarding laying off the CDOT marketing. Unfortunately I am compelled to add that to the discussion when people are purposely writing misleading information about CDOT and WAFL and other NetApp features. I do that because either people who write this stuff either don’t know what they are talking about or, more likely, they are writing it because they are employed by a competitor to NetApp.
Net is that if CDOT is nothing special and WAFL sucks so bad in performance as you say, why do tech companies like Cisco, Oracle, Amazon, Yahoo, Verizon Cloud, AT&T Cloud along with the US Government, the NFL and many others run their businesses on it? It’s not inexpensive, so there must be some good reason why these companies are spending their hard earned money on it. I doubt it is because CDOT does nothing for them and WAFL performance sucks.
Playing semantics to justify your lack of real tiering technology doesn't help your case, How does what a high level analysts opinion, even the founder of Storage Switzerland bolster your case ? As the self appointed founder of 'Storage Uzbekistan' I'm stating tiering and caching are different things.
If you'd like to argue the point further then please provide a referencable definition for both caching and tiering and then let's see which one really applies to flashcache. This corrupting of industry standard terminology to cover cracks in Netapp's feature list has been going on for years now, it's just people got sick of calling them out on it.
Pretty much of those organizations also run everyone else's storage kit, so what ?.
I thought that I did provide a referable definition for both caching and tiering. The two largest storage companies in the world, EMC and NetApp define SSD cache as a "Tier". Things like this are defined by industry leaders. If EMC and NetApp who are the two largest storage companies in the world and who rarely agree on anything both call SSD cache a tier, it is pretty difficult to argue that definition is incorrect. NetApp and EMC also sell more storage than all other storage vendors combined and actually develop and design their own products vs. companies like Dell and HP who OEM other storage companies (DotHill/NetApp) products or buy storage companies outright (LeftHand, Equallogic, Compellent, 3PAR) and sell that product as their own. SSD Cache and FlashCache provide the same benefit as physical auto tiering which is moving or copying data to faster technology to provide better performance without having to purchase all high performance disk even though only a subset of your data needs it at a given point in time.
Gartner also states in their report "SSDs Simply Can’t Replace Hard Drives" : What datacenter operators and server manufacturers alike already know is that the rise of in-memory databases are creating a new, all-DRAM tier of storage. SSDs are being inserted “underneath” them, creating a slightly less “hot” tier of frequently accessed data.
As the industry moves forward, all storage vendors will be using virtual tiering where they use SSD's or Flash Memory as performance acceleration in front of slow disk vs. having data sit on high performance disk. That, along with SSD prices coming down, is one reason 15K drives are going away. You can see this in real time as practically every startup is doing it this way as the two storage industry leaders EMC and NetApp. We can agree or disagree on what to call it caching or tiering, but net is physical auto tiering is going to be replaced by virtual tiering. What we will be left with is SSD drives and Flash memory augmenting performance of large NL-SAS drives where the data will actually be stored. This will be the way it works until SSD’s can replace all mechanical disk which won’t be anytime soon as it will be too costly to build Petabyte storage solutions with all SSD’s for the near term.
I guess to turn the table around, show me an industry storage leader or analyst that says that SSD Cache is not a tier. I've indicated three, EMC, NetApp and Gartner who say that it is.
Tiering is dead long live caching...er we meant tiering,..no caching...no tiering...
Talk about revisionist history, seems it wasn't dead afterall and Netapp are trying desperately to remain relevant.
NetApp says tiering is dying
http://www.theregister.co.uk/2010/02/19/netapp_tiering_dying/
NetApp slammed over “tiering is dead” comments
http://searchstorage.techtarget.com.au/news/2240019553/NetApp-slammed-over-tiering-is-dead-comments-EMC-savaged-by-IBM-and-Pillar
My personal favourite.at the time.
Netapp & tiering - Just when they thought things were looking up.
http://www.storagerap.com/2010/02/netapp-tiering-just-when-they-thought-things-were-looking-up.html
You're relying on some marketing blurb in a EMC FastCache document to try and justify your position. The simple difference here is that tiering uses persistent storage to move data to the most relevant place whereas, caching copies data to a temporary area for fast access. Flash cache (PAM) whatever you want to call it does the latter, which means it's a cache. If you want further proof it's also non persistent and hence why it can only handle reads.
That's the basics, but there are all kinds of nuances as to how they affect workloads, in plenty of cases that read cache will do a better job. So I'm not saying it's crap, but I am saying its still a read cache despite you marketing efforts, just like Netapp Flex Clones are really nothing more that a read writeable snapshot. Changing the terminology may reassure the marketing department and mislead the gullible but it doesn't make it true.
D@ Netapp seemed to think it was a cache, of course this was in the bad old days before Netapp decided tiering wasn't dead afterall :-)
"For instance, Megacaches can be used to dramatically change the I/O reaching the back-end disks of the array. NetApp FAS systems can have up to 16TB of deduplication-aware, ultra-granular (4K) and intelligent read cache."
http://recoverymonkey.org/2011/06/29/buyer-beware-is-your-storage-vendor-sizing-properly-for-performance-or-are-they-under-sizing-technologies-like-megacaching-and-autotiering/
"Pretty much of those organizations also run everyone else's storage kit, so what ?."
Um no, these companies don’t primarily run everyone else's storage kit. Cisco runs primarily on NetApp, Oracle actually develops Oracle on NetApp, Amazon Cloud runs primarily on NetApp, Yahoo e-mail all runs on NetApp, Verizon Cloud and AT&T Cloud run primarily on NetApp and the US Government runs 42% of its storage on NetApp and 27% of its storage on EMC. The vendors who do physical auto tiering HP, Hitachi and Dell make up 6.9%, 6.3% and 3.7% of US Federal Government storage share respectively, totaling 16.9% which is less than half of NetApp’s share of US Federal Government storage. If physical auto tiering is so awesome and virtual auto tiering “covers cracks in NetApp’s feature list” explain to me again why NetApp has so much market share, is growing and the companies that have physical auto tiering all added together don’t have as much share as NetApp and all three of them along with EMC lost market share in the full year of 2013, while NetApp gained. Say what you like, but the numbers tell the tale. Customers aren’t buying NetApp and moving to NetApp because it is crap.
Most actually use whatever comes cheapest / free. For instance Oracle may well do a large part of development on Netapp, but they also do have shed loads of everyone's storage. So again it comes down to semantics As for government contracts from a technology perspective they're meaningless, they're based buying frameworks with a 5 year cycle, your over reliance on US government is precisely why Netapp are suffering now.
Scale out clustering .... Because the only way to get any semblance of performance from WAFL is to source your volumes from separate aggregates on multiple HA pairs and stitch them in to a single name space. Nice, affordable solution that! I must buy me one, right after I strike gold.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
