Try this point of view on for size: there is no general large scale file storage problem. Companies set up to deal with that problem have failed to set the world on fire and over-invested ones, like ONSTor and Copan, are facing difficulties. Meanwhile, block storage SAN re-invention companies, such as 3PAR and Compellent have …
I agree, sort of....
My first response to the article, possibly because of the somewhat hard line opinions was to say this is all wrong. Actually on reflection, its probably not far off the mark but there are more factors involved here which werent discussed and distort things somewhat.
For instance, Centera had a major advantage over new startups because all Object stores require the application’s adoption of the store’s API. Would you modify your application to work for Centera and get a bit of free EMC marketing or modify your application with a complete unknown? No brainer really. CAS vendors are at the mercy of the application vendors. If however your application isnt that well known and you can get kudos for supporting a particular CAS then you will support that CAS. Case of the tail wagging the dog.
I certainly agree that the massively scaleable systems arent much use to most companies.
But also think that there is a general lack of adoption for new products isnt always because of lack of need for the products, but a reluctance to move from the norm. I think a perfect example is Data Domain. They make existing technologies better. Ie they de-duplicate data from legacy systems. Too me this is a band aid. But because its a band aid that makes legacy products perform better , its going to get positive press (or at least not negative press) from those vendors. What would be better of course is that you dont duplicate the data in the first place. Products that do this would save customer money. But those products are new paradigm and existing vendors cant compete and consequently the band aids get the press.
Another example are file redirectors (file virtualization ) . For me it seems to simplify a storage problem. Everyone points at one place and the system will get you the data. If you need to tier the storage , keep redundant data, track usage, NAS migration etc you can do this. But the problem is who the heck is going to promote this technology? Its not going to be the NAS vendors for sure as their storage just been commoditized. Consequently that technology is doomed to failure. (PS I am not a provider of such technology and in facts partly compete with what we do, but still it makes sense)
But the storage and data management industry is steered by the old guard storage companies and the storage analyst ,who largely voice their last client’s opinions or the big storage companies. (ps not all analyst , there are some who do provide an honest outlook).
So I am not saying that some of the solutions created for a market that didnt really exist but I think there is also prevailing attitude of screw what the customer needs, lets focus on the vendors.
The real storage problem ...
certainly at our place ... is old servers with a couple hundred gig space being filled up ...
whats the problem?
all the shared drives etc are all on these older servers, profiles mapped to them, server room full of servers with no space for more, yet no space to consolidate and make space, drive mappings all over the shop using up every spare GB of disk space.
Just put in a NAS? not quite so easy with profiles, drive mappings and the like for a couple hundred users spread out over 10 locations - its a smallish company with a lot of remote offices.
the time to sort it all out without considerable user inconvniance just isnt there.
Its the legacy of old servers, old disks thats the problem. Sure, theres a bit more data stored, but mostly thats just temp web traffic and pointless downloads rather than anything that needs saving on a server.
upgrades with down time, user inconveniance, no physical space and old servers ... thats the problem.
try asking your boss for 10k to sort it out in a recession! haha
right ... beer day. have fun folks!
Good see that you were able to see through the mist and see what's really important today - lost cost solutions. Thank you..
Having worked for one of the HPC storage companies for a year and a half, (Crosswalk, which failed for other reasons.) And in SANS at Mcdata, this seems to hit the mark. In crosswalk's situation we couldn't maintain a solid stable platform enough to satisfy our customers. The system worked well enough, provided the speeds our customers demanded, its just that it was built on a standard redhat linux system with little attention paid to the various packages installed, services running, etc. Those inhibited the filesystem performance and detracted from the product overall.
Our customers seemed generally excited by the performance, but yes it was entirely a niche product. Had there been enough success on the sales side we might have remained in business, but in today's market I doubt it. The virtualization product we had hoped to developed also had high hopes but never seemed to materialize.
As for storage being a problem, I don't think it is - many products offer search capabilities to find your file(s) for you or handle duplication. And people don't really care all that much that a file is duplicated.
I've built several systems on Linux using MD and LVM to create raid 1, 5, and 6. Yet when people can buy a 1TB external drive, or IBM, EMC, HDS can come in and just add more capacity - customers prefer that. Perhaps the industry expected or hoped to create demand where it didn't exist?
Having been at the sharp end of this market for a couple of years, I can wholly agree with Chris's comments. I am not sure that I wholly agree with some of the vendors or platforms he mentions, but the essence is 100% correct.
this whole area was a marketing conundrum where "there is a gap in the market, but no market in the gap". All end users saw the sense from a technology perspective but there was no real Business justification and that is where they all fall down. Many congratulations to Acopia therefore in palming off their flawed product set on F5 for $250m+ which even at the time of sale could not have been worth 20% of that in truth; what would they be worth now??!!
A Different View
According to IDC, file-based data will account for nearly 80 percent of all data within the enterprise by 2012. This seems to point to a very real challenge – how to cost-effectively store and manage such rapid file-based data growth?
Today’s traditional, scale-up solutions were simply not designed to answer this question. And, as of now, none of the major storage players have demonstrated an ability to engineer and bring to market a viable, enterprise-ready, scale-out NAS product. This appears evident when you review the rising cost of OPEX related to enterprise storage management.
Meanwhile, more than 1000 customers across the enterprise, including mainstream industries such as Manufacturing and Pharma, are using Isilon scale-out NAS because the growth of their file-based data had overwhelmed their traditional storage systems and Isilon was clearly the best solution to address their challenges going forward.
More than 250 storage start-ups followed Isilon into the marketplace in 2001. Regardless of market opportunity, this was never going to be sustainable. Today, market forces have winnowed the field down to the companies with the most competitive and value-driven offerings. With file-based data growth showing no signs of slowing, it seems clear the challenges inherent in skyrocketing stores of file-based data are making scale-out solutions more viable in the enterprise than ever before.
Finally, mainstream adoption typically follows “niche” markets who compose the first wave of adoption. Today we are witnessing the next wave, with the greatest uptick in adoption coming from mainstream enterprise businesses experiencing the same file-based data challenges the “niche” markets experienced a few years ago.
- Lucas Welch, Isilon Systems
@A Different View
No offence, but I fail to see the "Different View". Unless I'm mistaken but Islon is an expendable NAS. If anything it re-inforces what Chris was saying. Its more of the same, albeit with a different twist (and I am sure it is very very good).
Would quite like to hear the justification for "was clearly the best solution to address their challenges going forward". I am not disputing it, just want to know why/how/what. (chance for a good marketing rebuttal)
There are still other ways achieving similar goals though using commodity hardware. For instance, there is Parascale ( www.parascale.com ) who can scale out using inexpensive hardware. For the AC above that had the issue of a multitude of fileservers, you could copy the data from the servers to this one platform but keep all the existing shares the same (i presume you would still have to change the logins) and have them all point to virtual IPs on this box.
Then there are companies that are approaching the problem from a different angle and adding value to the management of the data. Its not just dumb storage, the storage does data management things. Companies like Tarmin ( www.tarmin.com ).
Lastly, we have other companies that take a more holistic look at data. Rather than it all being about just "storage" , they manage the "data" through the whole lifecycle. Cofo ( www.cofio.com ) have a product that start the data management form the primary and will move it down the food chain in a true ILM paradigm incorporating replication / backup / CDP and archiving. So rather than managing a whole bunch of disjointed products, you manager it through one seamlessly, even if the final destination is a different storage vendor.
So there are a quite a few companies that offer technology that are designed to reduce cost. By using commodity hardware, reducing the data retained by managing it intelligently, decreasing the man management burden etc.
Thanks for your thoughtful response. Your points are well taken and we at Isilon agree with your approach. Storage should not require you to change how applications run but should instead simplify data management and streamline operations for file-based applications. In fact, this very value proposition is why more than 1000 customers across a wide variety of enterprise industries selected Isilon IQ as their primary storage solution – Isilon solved their file-based data storage challenge by not forcing them to make a choice between performance/ capacity or applications/costs, delivering complete flexibility in managing file-based data growth, streamlining IT operations, and accelerating application performance.
From our perspective, our products key differentiator is ‘scale-out’. Unlike traditional, scale-up storage that requires a forklift upgrade to scale storage capacity/performance, scale-out storage allows you to scale by simply adding nodes to your cluster. It also means that as you scale your unit of management remains ONE – a single volume as opposed to multiple RAID groups, LUNs, and file system. The complexity of multiple LUNs, volumes, and so on created by traditional storage is a fundamental pain point our customers are trying to solve when replacing single-head RAID based system with Isilon IQ.
Via our operating system OneFS, Isilon provides a holistic data management solution combining industry-standard hardware components (Intel CPU, Ethernet, and SAS/SATA disks) with industry-leading data protection, replication, and capacity provisioning, providing a single, highly scalable, high performance, shared pool of storage, clearly differentiating Isilon from other “scale-out” systems. In short you don’t need to change you application environment to take advantage of a much simpler, flexible, and more scalable storage architecture that reduces operating expenses and increases workflow productivity.
- Does Apple's iOS 7 make you physically SICK? Try swallowing version 7.1
- Fee fie Firefox: Mozilla's lawyers probe Dell over browser install charge
- Pics Indestructible Death Stars blow up planets with glowing KILL RAY
- Hands on Satisfy my scroll: El Reg gets claws on Windows 8.1 spring update
- Video Snowden: You can't trust SPOOKS with your DATA